UNIT 20: Case Study

MBCQ-721D

ES
UP
Quantitative Techniques
for Management
E-
Applications
CC
(c)
 Quantitative Techniques for Management Applications

Course Design

S
Advisory Council

E
Chairman
Dr Parag Diwan

Members
Dr Anirban Sengupta Dr Ashish Bhardwaj

UP
Dr Kamal Bansal
Dean Dean CIO

Dr S R Das Dr Sanjay Mittal Prof V K Nangia

VP – Academic Affairs Professor – IIT Kanpur IIT Roorkee

SLM Development Team

Wg Cdr P K Gupta
Dr Joji Rao
Dr Neeraj Anand
E-
Dr K K Pandey

Print Production

Mr Kapil Mehra Mr A N Sinha

CC

Manager – Material Sr Manager – Printing

Author

S Jaisankar

All rights reserved. No part of this work may be reproduced in any form, by mimeograph or any other means,
without permission in writing from MPower Applied Learning Enterprise.
(c)

Course Code: MBCQ-721D

Course Name: Quantitative Techniques for Management Applications

Version: July 2013

© MPower Applied Learning Enterprise

UNIT 20: Case Study

S
Contents

E
Block-I

Unit 1 Decision Making ............................................................................................................. 3

Unit 2 Functions....................................................................................................................... 15

UP
Unit 3 Equations ...................................................................................................................... 23
Unit 4 Applications of Linear Equations and Functions in Business ................................... 35
Unit 5 Case Studies.................................................................................................................. 53

Block-II

Unit 6 Matrices......................................................................................................................... 57
Unit 7 Determinants ................................................................................................................ 83
E-
Unit 8 Probability................................................................................................................... 105
Unit 9 Random Variables and Probability Distributions..................................................... 123
Unit 10 Case Studies................................................................................................................ 153

Block-III
CC

Unit 11 Decision Theory .......................................................................................................... 159

Unit 12 Linear Programming .................................................................................................. 167
Unit 13 Transportation Models ............................................................................................... 185
Unit 14 Assignment Problem .................................................................................................. 199
Unit 15 Case Studies................................................................................................................ 207

Block-IV

Unit 16 Game Theory............................................................................................................... 213

(c)

Unit 17 Markovian Model........................................................................................................ 223

Unit 18 Data Collection ........................................................................................................... 229
Unit 19 Presentation of Data................................................................................................... 255
Unit 20 Case Studies................................................................................................................ 261
 Quantitative Techniques for Management Applications
iv

Block-V

S
Unit 21 Sampling ..................................................................................................................... 265

Unit 22 Basic Tools of Data Analysis–I .................................................................................. 287

Unit 23 Basic Tools of Data Analysis–II ................................................................................. 301

E
Unit 24 Forecasting.................................................................................................................. 309

Unit 25 Case Studies................................................................................................................ 371

Glossary ............................................................................................................................................ 375

UP
E-
CC
(c)
UNIT 1: Decision Making

S
Notes

___________________

___________________

E
___________________

___________________

___________________

___________________

UP
___________________

___________________

___________________

___________________

BLOCK-I
E-
CC
(c)
Detailed Contents Quantitative Techniques for Management Applications

S
Notes
UNIT 1: DECISION MAKING
___________________ UNIT 4: APPLICATIONS OF LINEAR EQUATIONS
AND FUNCTIONS IN BUSINESS
z Introduction
___________________ z Introduction
z Managerial Decision Making System
___________________ z Supply and Demand Functions

E
z Managerial Decision Making Environment
___________________ z Irregular, Unequal and Discontinuous Functions
z Quantitative Models
___________________ z Quadratic Equations
UNIT 2: FUNCTIONS
___________________ z Fitting a Quadratic Cost Curve
z Introduction

UP
z Functions
___________________ UNIT 5: CASE STUDIES
z Variables
___________________
z Types of Functions
___________________
UNIT 3: EQUATIONS
___________________
z Introduction
z Linear Equations
E-
CC
(c)
UNIT 1: Decision Making

Unit 1
3

S
Activity
Notes
Define the term decision.

Decision Making
___________________

___________________

E
___________________
Objectives
___________________
After completion of this unit, the students will be aware of the following
topics: ___________________

\ Quantitative decisions ___________________

UP
\ Importance of variables ___________________
\ Importance of eight decision making tools
___________________

___________________
Introduction
___________________
A decision always involves choice among several alternatives. In
the most basic sense a decision always involves the answer to the
question "to do or not to do?" Not to do (inaction) determines that
decision. To do (action) usually involves different options. The
E-
mathematical model identifies the optimal way, but for a variety of
reasons, other satisfying options may be selected and acted upon.
There are industry-wide and market-wide decisions that have to be
made. Often these decisions must transcend domestic
considerations to incorporate international aspects.
CC

Managerial Decision Making System

Every decision making task results in an output which is the
evidence of the decision taken. In industry, it is ultimately some
kind of product, that is, a good service or on idea. The reasoning
takes place in the decision making rectangle which is sometimes
referred to as, quite appropriately, the black box. Here a
transformation of the inputs takes place that results in the output.
The transformation process has both physical and mental
properties. On the input side a large number of variables may be
listed. These variables can be classified in terms of the traditional
factors of production, i.e., land, labour and capital as well as the
(c)

more recently emerged complex variables related to systems,

technology and entrepreneurship. Underlying this input-output
system is a feedback loop identified as managerial control system.
Its function is to optimize the transformation of inputs into the
desired output. Seen in a nutshell, in industry, optimization means
 Quantitative Techniques for Management Applications

4 the minimization of costs and the maximization of profits subject to

S
Notes
Activity legal, social and ideological constraints.
What is Managerial Decision
___________________
Making System?
The computer has forced the decision maker to very carefully
___________________ delineate and quantify the variables that makeup the building
___________________ blocks of the decision task. What is needed and how much is

E
needed for decision optimization have become the important
___________________
questions. In addition, the proper time sequencing of the decision
___________________ variables within the decision process had to be understood. And all
___________________ answers had to be unequivocally quantified. It soon became

UP
apparent to every decision maker that quantified variables had
___________________
different properties and specific quantitative control mechanisms
___________________
had to be designed. Not only was the decision maker confronted
___________________ with variable-inherent properties, the decision tasks themselves
___________________ have such peculiar quantitative properties.

A variable, the building block of the decision task, may be seen as a

small piece of a complex behaviour. Buying a house,
manufacturing a product, spending money on a show are examples
of variables. Each variable represents a distinct dimension of the
E-
decision making task. So the decision space is always
multidimensional, and it is a major task for the decision maker to
find out which variables make up that space. If an important
variable is overlooked, obviously the decision will be less than
optimal. Furthermore, the quantitative impact of the variable
must be ascertained. And here the special variable-inherent
properties come into play. The following examples may show the
CC

differences among the three types of variables.

Deterministic variables can be measured with certainty. Thus,

equal measures have equal cumulative impact, or, to use a simple
example, a+a = 2a.

Stochastic variables are characterized by uncertainty. Thus,

a+a=2a+X, where X is a value that comes about because of the
uncertainty that is associated with the variable.

Heuristic variables are those that exist in highly complex,

unstructured, perhaps unknown decision making situations. The
impact of each variable may be explained contingent upon the
(c)

existence of a certain environment. For example,

a + a = 3a but only if certain conditions hold. Actual industrial
decision making situations in each case may involve the number of
gallons of aviation fuel obtained by cracking a barrel of crude oil
(deterministic), projected product sales given amount spent on
UNIT 1: Decision Making

advertising the product (stochastic) and the construction of a 5

S
platform in outer space (heuristic). Notes
Activity
What is the reason for the
Check Your Progress ___________________
existence of a managerial
hierarchy?
Fill in the blanks: ___________________

E
___________________
1. Reasoning is ……………….. by nature, which can be
rational or irrational. ___________________

2. All types of human decision-making are essentially ___________________

……………… processes. ___________________

UP
3. The two most common flaws in decision making are ___________________
inertia and……………….. . ___________________

Managerial Decision Making Environment ___________________

___________________
The reason for the existence of a managerial hierarchy, that is,
lower, middle and top management, finds itself in different
parameters in which an organization operates. There are industry-
wide and market-wide decisions that have to be made. Often these
decisions must transcend domestic considerations to incorporate
E-
international aspects. Such decisions—usually made by top
management—occur in a broad-based, complex, ill-defined and
non-repetitive problem situation. Middle management usually
addresses itself to company-wide problems. It sees to it that the
objectives and policies of the organization are properly
implemented and that operations are conducted in such a way that
CC

optimization may occur.

You may note that while most of the quantitative decision making
tools —indeed virtually all of the deterministic tools—were
developed to optimize the decision making process, actual
managerial practice has sometimes moved away from that
objective. The previously mentioned legal or social constraints
often at times do not permit optimization and satisfying has been
substituted for it. Satisfying refers to the attainment of certain
minimum objectives. For example, a company may have the
economic and technological power to smother the competition
within its industry but refrains from doing so because of MRTP
(c)

considerations. Big size per se may be considered in violation of the

law or in the international arena, may result in the imposition of
quotas.
Lower management is responsible for the conduct of operations—
the firing line so to speak is this—in production, marketing,
 Quantitative Techniques for Management Applications

6 finance or any of the staff functions like personnel or research.

S
Notes This decision environment is usually well-defined and repetitive.
___________________ Obviously, with reference to a given decision making situation, the
distinction between top, middle and lower management may
___________________
become blurred. In other words, in any ongoing business there is
___________________

E
always a certain overlapping of the managerial decision making
___________________ parameters.
___________________ The study and analysis of the existence and interaction of these
___________________ parameters is of great importance to the management systems

UP
designer or communication expert. From the quantitative
___________________
managerial decision making point of view, their importance lies in
___________________
recognizing their peculiar constraints and then to build the
___________________ appropriate decision models and to select the best suited
___________________ quantitative decision tools. A brief discussion of each environment
in this light may enhance the understanding of the tools that are
discussed later on. The company’s approach to the domestic or
international market is filtered through industry-wide
considerations. What does the market want, what does the
E-
competition already supply? Where is our field of attack? Do we
have the know-how, do we have the resources? What is the impact
of our actions upon the market, our own industry and other
industries? These are some of the questions that have to be asked,
defined and answered. The problems are unstructured and
complex. Thus, often a heuristic decision making process can be
utilized to good advantage. Forecasting is of major importance and
CC

hence stochastic decision making is widely employed in this

uncertain decision environment. But even a deterministic tool—
usually intended for decision making situations that assume
certainty—input-output analysis, can be effectively used in this
environment.
Middle management decisions are primarily company-wide in
nature. As mentioned before, these decisions steer the organization
through its life cycle.
Major features of a firm’s life are objectives, planning, operation
and the ultimate dissolution. The objectives are general and
(c)

specific in nature. Obviously, the top management establishes the

objectives, but middle management functions as their guardian.
Indeed, every decision at this level must provide feedback control
for each of the other components.
UNIT 1: Decision Making

7
Planning refers to both policy execution as well as policy

S
Notes
development. Scale of production, pricing of the product, product
mix, in short the orderly and efficient arrangement of the input ___________________
factors is to be decided at this point. Making these factors into a ___________________
product is the job of operations. Some operations have been

E
___________________
traditionally called line (financing, production, and marketing) and
others staff (personnel, research, etc.); yet, in the quantitative ___________________

decision systems of the modern firm, such differences are difficult ___________________
to trace in the decision patterns, because the same decision, ___________________

UP
making tools are employed. Since the decision environment at this
___________________
level is somewhat more structured than at the top level but still
highly uncertain, stochastic decision tools are frequently employed. ___________________
In those finance, production and marketing situations that can be ___________________
well-defined, may be repetitive, deterministic decision tools are
___________________
found.
It may appear somewhat odd that the decision environment
includes attention being paid to the dissolution of the firm. The life
cycle concept has been mentioned, and it will be encountered again
E-
as one of the major underlying conceptual aids in forecasting. It is
well known that business organizations are born, live and die like
natural organisms.
Therefore, decision making should always be cognizant of the
possibility of dissolution. That moment comes when, to use the
vernacular, good money is thrown after bad. While market forces
CC

and the application of quantitative analysis normally show the

approaching occurrence of that moment—even if the management
involved shuts its eyes to the facts or is ignorant about them—at
this point the decision is made or superimposed to opt for a
turnaround or dissolution. Public agencies unfortunately are rarely
subject to such stress producing alternatives.
The lower management decision making environment represents a
specialized, narrowly defined area within a company’s total
decision or operational field. Supervisory personnel of all types are
operating in this environment. The decision tasks are normally
well defined and repetitive. While the element of uncertainty never
(c)

leaves the decision environment, here uncertainty can often be

programmed into general or sub-routine, and stochastic decisions
taken as if they were deterministic in nature. A good example is
the pricing system of clothing discounters. Merchandise is put on
the floor at price A on day one. On, say, day ten the price is
 Quantitative Techniques for Management Applications

8 automatically reduced to price B and so on until the article is

S
Notes
Activity either sold or given to charity after thirty days. This is known as
What is Problem Definition?
___________________ programmed decision making. It should be noted that while the
nature of the decision environment remains intact, the decision
___________________
maker’s tasks have been greatly reduced. The complex variables
___________________

E
and unstructured decision environment of the merchandising task
___________________ have been placed first into a model and then into decision making
___________________ sequence (algorithm). This is the general idea behind model
building and the development of algorithms.
___________________

UP
It is highly important that every decision maker has a firm
___________________
understanding of the philosophy upon which quantitative decision
___________________
making is based. Under no circumstances is it sufficient to just
___________________ know how to perform a certain quantitative analysis and to obtain
___________________ a solution to be able to make a decision.
To turn to the specific aspects of the quantitative decision making
process, it is possible to recognize three distinct phases in every
decision situation. Given a carefully defined problem, a conceptual
model is generated first. This is followed by the selection of the
E-
appropriate quantitative model that may lead to a solution. Lastly,
a specific algorithm is selected. Algorithms are the orderly
delineated sequences of mathematical operations that lead to a
solution given the quantitative model that is to be used. The
algorithms generate the decision which is subsequently
implemented by managerial action programs.
CC

Problem Definition
Problem definition is a cultural artifact which is especially visible
in a society’s economic and industrial decision making process.
Obviously, if such cultural determinants are operative in the first
phase of managerial decision making, their effect can be noticed at
various stages in the process irrespective of the quantitative, thus
hopefully objective, methods that are used in the design of the
models and algorithms as well as the decision itself.
A brief digression into problem identification may be in order at
this point. For purposes of this book and for quantitative
(c)

management decision methodology in general, it is presupposed

that a problem has been identified.
In the private sectors of free enterprise economies, however, a
manager’s ability to recognize problems and even to anticipate
problems that may emerge at some future time is vital to the
UNIT 1: Decision Making

survival of the firm. Those managers that make effective decision 9

S
concerning a known problem are good administrators; those that in Notes
Activity
addition can recognize and anticipate problems are creative. It is What does conceptual model
___________________
known that creativity is partially inborn and partially acquired. represents?
___________________
Thus, the quantitative decision maker will not only try to master

E
the methodology but also attempt to sharpen his or her problem ___________________

identification skills—his or her creativity. ___________________

___________________
Check Your Progress
___________________
Fill in the blanks

UP
___________________
1. Inertia is often due to a fear of…………….
___________________
2. Problem definition is a cultural artefact which is
especially visible in a society's economic and ___________________

industrial……………… ___________________

The Design of Conceptual Models

The conceptual model represents the logic that underlies a
decision. Based on this logic the quantitative model and specific
E-
algorithms are constructed. The logic may be a priori or empirical
in nature, e.g., when shooting craps in a casino, a gambler has pre-
established a conceptual model concerning the odds of the game.
On a priori ground—using only his or her intellect—in determining
the odds of every roll of the dice, the concept dictates that the win
of a seven or eleven on the first roll has likelihoods of 6/36 and 2/36 ,
CC

respectively. (There are 6 possible combinations of spots showing

on 2 dice that yield a seven and 2 combinations that yield an
eleven with 36 combinations for all spots from two through twelve.)
Given this conceptual model, quantitative models and algorithms
can be designed that facilitate the betting decision.
Now suppose that our gambler stumbles across a floating craps
game in some dark alley. After observing the action on the
pavement for a while, he notices that seven’s and eleven’s do not
occur on the first roll with the likelihood dictated by his conceptual
model. Rather there seems to be a preponderance of two’s, three’s
or twelve’s—which he knows are losses. Crooked dice, he may very
(c)

quietly think to himself. For crooked dice, an a priori logic which is

based on the ideal situation in which every spot on a dice has an
equal probability of occurring (1/6) and any spot on two dice as well
(1 / 6 × 1
/6 = 1
/36) according to the multiplication theorem) is
unsuitable. Rather, he will now ascertain by observation (by
 Quantitative Techniques for Management Applications

10 experiment) the empirical probabilities which are determined by

S
Notes the weights that have been cleverly or crudely (it is a dark alley)
___________________ concealed in or on the dice. Once this empirical conceptual model
has been generated, our gambler may continue the betting decision
___________________
process in terms of the amount of the bets at each roll, etc. He may
___________________

E
also redefine the problem and leave.
___________________
Conceptual models may take many forms. In every case the
___________________ general design intent is to understand and to depict the reality
___________________ that relates to the problem. Most conceptual models show a

UP
functional relationship in graphic or matrix format. All models
___________________
that are used in this book are of this type. But it is also possible,
___________________
indeed necessary in some decision cases, to build a physical model.
___________________ In the natural and engineering sciences, it is the usual form. If the
___________________ decision involves mass production of some item, the physical model
is known as a prototype.
In the design of the conceptual model, it is important to observe
that the decision maker clearly delineates the interrelationships
that make up the reality—or the systems—in which the problem
E-
occurs. But in the model building process it is virtually impossible
to include all variables that have a bearing on the decision. The
model includes only the major variables (endogenous variables) as
seen from the decision maker’s vantage point. There will be always
decision-related variables that exist outside of the decision space
(exogenous variables) because of their unrecognized status or
CC

conscious exclusion due to time, cost or limited impact

considerations. Such variables should be kept mentally ready
because over a set decision horizon they may indeed become
sufficiently important to be included into the system.
Once the conceptual model has been designed and its logic
expressed in terms of some systems configuration such as the
graph or matrix or perhaps network or flow diagram, the
quantitative models are simply superimposed by quantifying the
logic. Once that has been accomplished a relatively minor task
remains in the selection of the algorithms and the computerization
of the process. Many a decision process has been commenced
(c)

needlessly and most of time injuriously, to some extent, because of

faulty problem definition or poor conceptual model-building. Then
there is no optimal or even satisfying outcome. To put it simply,
number crunching and possible error correction is relatively easy,
even though the reader may not immediately share this view as he
UNIT 1: Decision Making

or she does just that in the units that follow. Only the difficult 11

S
tasks, that is, sound insight into the problem and its careful Notes
Activity
definition as well as proper logic employed in the conceptual model What is transition?
___________________
building process, will yield sound decisions and outcomes. Here
___________________
errors are very difficult to correct.

E
___________________
Quantitative Models ___________________

Once the conceptual model has been properly designed, the ___________________
quantitative model and its algorithms should almost “flow” out of ___________________

UP
it. The transition is natural, smooth, and almost automatic. The
___________________
quantitative model is selected from the many such models that
have been designed by mathematicians. So while the decision ___________________
maker will always build a conceptual model, the quantitative
___________________
model is typically selected from an available pool of such decision
making tools. The selection is made on the basis of the ___________________

predominantly stochastic, deterministic or heuristic nature of the

variables. There are available quantitative models for each kind as
discussed in the following units, and the decision makers task is to
select the appropriate one for a given decision situation. “Know thy
E-
tools” should be inscribed on every decision maker’s desk. As it is
possible to build a wall with a spade when the trowel would be the
more appropriate tool, decision makers may sometimes misuse
quantitative tools.

The Decision
CC

A decision always involves choice among several alternatives. In

the most basic sense a decision always involves the answer to the
question “to do or not to do?” Not to do (inaction) determines that
decision. To do (action) usually involves different options. The
mathematical model identifies the optimal way, but for a variety of
reasons, other satisfying options may be selected and acted upon.
These other options are firmly rooted in an organization’s
objectives and planning activities. As shown in greater detail later,
a decision maker always has control over setting the objective and
planning which interfaces with policies, strategies and tactics. But
one has no control over the reaction to the decision within the
market environment. Here various states, collectively known as
(c)

the states of nature, emanating from customers, suppliers,

competitors, public agencies, etc., render the final judgment about
the soundness of the decision. The decision is the end product of a
sequence of mental activities as illustrated in the preceding pages.
To make a decision does not necessarily mean that it gets carried
out. In order to accomplish that, numerous managerial action
 Quantitative Techniques for Management Applications

12
programs are necessary. They represent the physical extension to

S
Notes the decision making process. This book stops at the point when the
___________________ decision is rendered. The action programs, the physical component,
cannot be discussed because they must be specifically designed for
___________________
each situation. A good decision maker, however, will try to place
___________________ the seeds for proper implementation into the decision.

E
___________________
Check Your Progress
___________________
Fill in the blanks:
___________________
1. A decision always involves choice among several

UP
___________________
____________ .
___________________
2. The decision is the ____________ of a sequence of mental
___________________ activities.
___________________
Summary
Every decision making task results in an output which is the
evidence of the decision taken. A variable, the building block of the
decision task, may be seen as a small piece of a complex behaviour.
E-
Buying a house, manufacturing a product, spending money on a
show are examples of variables.
The reason for the existence of a managerial hierarchy, that is,
lower, middle and top management, finds itself in different
parameters in which an organization operates. There are industry-
wide and market-wide decisions that have to be made.
CC

The conceptual model represents the logic that underlies a

decision. Based on this logic the quantitative model and specific
algorithms are constructed. Once the conceptual model has been
properly designed, the quantitative model and its algorithms
should almost "flow" out of it. The transition is natural, smooth,
and almost automatic.
The mathematical model identifies the optimal way, but for a
variety of reasons, other satisfying options may be selected and
acted upon. These other options are firmly rooted in an
organization's objectives and planning activities.
(c)

Lesson End Activity

“A decision always involves choice among several alternatives”.
Comment.
UNIT 1: Decision Making

Keywords 13

S
Notes
Deterministic Variables: Deterministic variables can be
___________________
measured with certainty.
___________________
Stochastic Variables: Stochastic variables are characterized by

E
uncertainty. ___________________

___________________
Heuristic Variables: Heuristic variables are those that exist in
highly complex, unstructured, perhaps unknown decision making ___________________
situations. ___________________

UP
Questions for Discussion ___________________

___________________
1. Every decision making task results in an output which is the
evidence of the decision taken. ___________________

2. Define variable. What are the different types of variables? ___________________

3. What are the key decisions taken at different managerial

levels?
4. Write a note on problem definition.
E-
5. Discuss the process of designing conceptual and quantitative
models.

Further Readings
Books
CC

R S Bhardwaj, Mathematics for Economics and Business, Excel

Books, New Delhi, 2005

D C Sanchethi and V K Kapoor, Business Mathematics

Sivayya and Sathya Rao, An Introduction to Business Mathematics

Web Readings
www.managementstudyguide.com

www.textbooksonline.tn.nic.in

www.mathbusiness.com
(c)
 Quantitative Techniques for Management Applications

14

S
Notes

___________________

___________________

___________________

E
___________________

___________________

___________________

UP
___________________

___________________

___________________

___________________
E-
CC
(c)
UNIT 2: Functions

Unit 2
15

S
Notes

Functions
___________________

___________________

E
___________________
Objectives
___________________
After completion of this unit, the students will be aware of the following
topics: ___________________

\ Functions ___________________

UP
\ Variables ___________________
\ Types of Functions
___________________

___________________
Introduction
___________________
When we say that quantity demanded of a commodity is a function
of its price, we mean that the quantity demanded depends upon
the price and, therefore, for a given price, we can determine the
quantity demanded. Here quantity demanded is a dependent
E-
variable while price is an independent variable. The behaviour of
the dependent variable with respect to change in independent
variable(s) is different in different situations. This fact gives rise to
several types of functions. A function f from a set X (domain) to a
set Y, a subset of X Y, is a rule that associates a unique element
of set Y (target) to each element in X.
CC

In this unit, we will study the functions and different types of

functions.

Functions
When we are talking about functions, we are not talking about
marriage ceremonies or birthday parties but are talking about
certain types of quantitative relationships between different
variables mathematically. For example, sales revenue is a function
of items sold and their price. We all know that. If we express it in
function form it would look like this:
(c)

Sales Revenue = f (No. of items sold) (Function Form)

Sales Revenue = f (Price per item) (Function Form)

 Quantitative Techniques for Management Applications

16
Function form only tells us that there is a relationship between the

S
Notes
Activity
variables. Here sales revenue, items sold and price per item are
What are functions?
___________________
variables.
___________________
Check Your Progress
___________________

E
Fill in the blanks:
___________________
1. Function form only tells us that there is a ___________
___________________
between the variables.
___________________
2. Sales ___________ is a function of items sold and their

UP
___________________
price.
___________________

___________________ Variables
___________________
Variables are the terms used for mathematical quantities that can
assume any values within a given set. The set of values of the
variable is known as the domain of the variable, which could be
limited (as is the case with water temperature which can vary
E-
between 0oC to 100oC) or can be unlimited (as distance).

Variables can be independent or dependent. Independent variables

are those variables whose values are not governed by the values of
another variable. In the case above, number of items sold and price
per item are independent variables. Variables whose values are
dependent on the values taken by another variable are called the
CC

dependent variables, e.g. sales revenue above.

In mathematics we say that whenever any variable Y is dependent

on another variable X for its values then variable Y is a function of
variable X. This means that whenever the value of variable X
would change then there would be a corresponding change in the
value of variable Y. Mathematically this is denoted as:

Y = f (X)

Now this relation between X and Y can take up any form, e.g., let
us say Y is double of X. This means when X takes the value 2 than
(c)

Y takes the value 2 × 2 = 4. So in the above function X is called the

independent variable and Y is called the dependent variable.

Note that the function does not specify the exact relationship
between X and Y, it only tells you that a relationship exists. The
exact relationship is defined by equation, something that we will
UNIT 2: Functions

see later. As stated above, the set of values of independent variable 17

S
is called the domain. The set of values of dependent variable is Notes
Activity
called the range. What is share price?
___________________

In the problems of business applications, generally the range and ___________________

domain will be real numbers.

E
___________________

How do we apply these functions and equations in the business? ___________________

The sales revenue example gives us a clue. Most of the activities in
___________________
the business are dependent upon some other events happening.
___________________
For example, if we want to know how the share prices move, we

UP
have to understand that the stock prices are dependent upon the ___________________
economic situation, industry conditions and company performance. ___________________
So the share price would be a function of all these factors and the ___________________
function would look like the following:
___________________
Share Price = f (economy, industry situation, company
performance)
Therefore, we can use functions to define any relationships that
exist between different processes and which can be measured
E-
mathematically. These models can be applied in making decisions
when the business situations are complex. An important point to
remember here is that, the functions do not imply a simple direct
relationship.
The applications of these models are manifold and as diverse as
the business situations we are in. The biggest benefit of using
CC

functions and equations for defining the relationships is that we

understand how exactly the processes work and what are the
constraints we are dealing with. For example, when we define the
production as a function of number of machines and time and we
know that we can produce a maximum of 500 units of a product per
hour per machine. So this becomes our constraint and helps us in
planning and scheduling. When we know that we need 5000 units
of that product in 2 hours, we can either use five machines (so that
each machine gives us 1000 pieces in 2 hours) or we get the work
done from outside. Of course, this is a very simple situation and we
will learn much more complex methods of mathematical
(c)

treatments of solving complex situations.

Writing the relationship in function and equation form above,
makes it easy for us to understand how the values of the
dependent variables change because of a change in any of the
independent variables on which it is based.
 Quantitative Techniques for Management Applications

18
Most of the quantitative methods we will be dealing with in this

S
Notes
Activity book, would make use of mathematical models, some of which we
Identify different types of
___________________
will develop as we go along. We will only take the basic
functions.
___________________ mathematical concepts as granted. Rest of the concepts would be
___________________ dealt with as if we are talking to a novice.

E
___________________ Check Your Progress
___________________ Fill in the blanks
___________________
1. We can use functions to define any __________ that exist

UP
___________________ between different processes and which can be measured
___________________ mathematically.

___________________ 2. Variables can be __________ or_____________

___________________
Types of Functions
We said earlier that the mathematical models of the functions can
take up different forms but we did not discuss the forms there.
Now let us understand the different forms that these mathematical
E-
functions can take up.

1. Constant Function: Let A denote a fixed number. Consider

that the function X has this value A for the value of X. So this
constant function can be denoted by:

f (X) = A for all X

CC

or more briefly by f (X) = A and we call it the constant function

A

Therefore, there can be many constant functions.

2. Identity Function: The function that associates to each

number X. The same number X is called the identity function.
This is denoted by:

Y = f (X) = X for all X

3. Linear Function: Consider the expression f(Y) Y = 2X+ 10.

For each number A, this associates the number 2A +10 to Y,
(c)

got by substituting A for X. If X = 2 then the value of Y is 4+10

= 14. This function tells us about the relationship between the
two variables X and Y.

4. Quadratic/Polynomial Function: Consider the expression

f(Y)X2 + 2X + 10. For each number A, this associates the
UNIT 2: Functions

number A2 + 2A + 10 to Y. This is a polynomial function and is 19

S
used for solving complex situations. Notes

___________________
5. Exponential Function: The function that associates the
number ex to each real number x is called the exponential ___________________
function. The properties of this function are given below:

E
___________________

ex+y = ex × ey ___________________

e-x = 1/ex ___________________

elog x = x ___________________

UP
1 1 ___________________
Here e= 1+ 2 + 2
+ ..........
___________________
x x2 x3
and ex = 1 + 1
+ 2
+ 3
...... ___________________
e is always positive.
x
___________________

6. Logarithmic Function: Log X is that number Y such that ey

= x. Log x is defined only when x is positive. The function that
associates log x to x is called the logarithmic function. Its
domain is the set of positive real numbers. Log x=y implies
E-
loge x=y and real as ‘log x’ on base ‘e’ is equal to ‘y’ which
implies ey = x but in frequent reading and writing we tend to
take the base as understood. But if we are using two different
bases then it must be mentioned. The two popular bases are ‘e’
and ‘10’. The properties of log function are given below:

Log 1 = 0
CC

Log (xy)= Log x + Log y

7. Modulus Function: For each real number X let |X| denote

the absolute value of X. It is also known as mod X.

|X|= X if X ≥ 0, and

|X|= -X if X < 0

For example |4| = 4 and |-4| = –(-4) = 4

These are only some of the examples of the different type of

functions that can be there. These functions are translated into
(c)

equations for use so that exact relationships are defined.

 Quantitative Techniques for Management Applications

20
Check Your Progress

S
Notes
Fill in the blanks:
___________________

___________________
1. The function that associates the number ex to each real
number x is called the ____________ .
___________________

E
2. The function that associates log x to x is called the
___________________
____________ .
___________________

___________________ Summary

UP
___________________
A function f from a set X (domain) to a set Y, a subset of X × Y, is a
___________________ rule that associates a unique element of set Y (target) to each
___________________ element in X. The unique element y in Y corresponding to an
element x in X is denoted as f(x) and is called the image of x. In
___________________
mathematics we say that whenever any variable Y is dependent on
another variable X for its values then variable Y is a function of
variable X. This means that whenever the value of variable X
would change then there would be a corresponding change in the
E-
value of variable Y.

Lesson End Activity

When we say that quantity demanded of a commodity is a function
of its price, what does it mean?
CC

Keywords
Function: A function f from a set X (domain) to a set Y, a subset of
X × Y, is a rule that associates a unique element of set Y (target) to
each element in X.

Exponential Function: The function that associates the number

ex to each real value of x is called exponential function.

Logarithmic Function: The function that associates log x to x is

called the logarithmic function.
(c)

Questions for Discussion

1. Define functions. What are the different types of functions?
2. Write a note on supply and demand functions.
UNIT 2: Functions

Further Readings 21

S
Notes
Books ___________________

R S Bhardwaj, Mathematics for Economics and Business, Excel ___________________

Books, New Delhi, 2005

E
___________________
D C Sanchethi and V K Kapoor, Business Mathematics
___________________
Sivayya and Sathya Rao, An Introduction to Business Mathematics
___________________

Web Readings ___________________

UP
www.textbooksonline.tn.nic.in ___________________

___________________
www.mathbusiness.com
___________________

___________________
E-
CC
(c)
 Quantitative Techniques for Management Applications

22

S
Notes

___________________

___________________

___________________

E
___________________

___________________

___________________

UP
___________________

___________________

___________________

___________________
E-
CC
(c)
UNIT 3: Equations

Unit 3
23

S
Notes
Activity

Equations
What does equation convey to
___________________
us?
___________________

E
___________________
Objectives
___________________
After completion of this unit, the students will be aware of the following
topics: ___________________

\ Equations ___________________

UP
\ Linear equations ___________________
\ Graphical solution
___________________

___________________
Introduction
___________________
Theory of equations is frequently used in solving the problems of
business. A statement of equality containing one or more variable
is known as an equations. Two or more equations are said to be
equivalent if they have the same solution.
E-
While the functions tell us that a relationship exists, equations give
us the exact relationship between the variables. These equations
take many forms and have one or more variables. For example, we
can say that Sales Revenue y = Number of items sold NX price per
item P. This is an equation with three variables Y, N, P and defines
an exact relationship amongst them. Here sales revenue Y is a
CC

dependent variable and other two are independent variables.

Here, we will discuss only linear and quadratic forms of equations.

Linear Equations
Linear Equation may be defined as an equation where the power of
the variable(s) is one, and no cross or product terms are present.
The general expressions of these linear equations look like the
following:
AX + B = 0
(c)

Here X = independent variable, A and B are numeric coefficients

This definition is a working definition.
Note that it is an accepted convention in mathematics
that letters from the beginning of the alphabet are used to typify
 Quantitative Techniques for Management Applications

24 known quantities and letters from the end of the alphabet are used

S
Notes to represent unknown quantities. In a linear equation, A and B are
___________________ real numbers which can be either positive or negative and may
involve fractions or decimals; it is also possible that B can be zero
___________________
but A cannot be zero for then it is not an equation and B also has
___________________

E
to be zero.
___________________
Example 3.1
___________________
A dealer sells a table and seven chairs for ` 6,000. The price of the
___________________ table is known to be ` 2,500, what is the price of one chair?

UP
___________________
Solution:
___________________
If X represents the price of each chair in units of ` 1 then we can
___________________
say that
___________________ 7 chairs × (Cost of each chair X) + Cost of table (` 2,500) =
` 6,000
` 7´X + ` 2,500 = ` 6,000, or simply
7X + 2,500 = 6,000
E-
7X - 6,000 + 2,500 = 0
7X - 3,500 = 0 (i.e., the form AX + B = 0)
7X = 3,500
X = 3,500 / 7 = 500

This means that each chair costs ` 500.

CC

The number of variables in a linear equation can be one or more

than one. For example, if

Y = AX + B

This equation can be rewritten as:

Y - AX - B = 0

That is the same form as we discussed earlier but it is an equation

with two variables.

The purpose of this form of the linear equation, with two variables,
is not to enable the problems to be solved but to state the
(c)

relationship between Y and X. The value of X can vary from time to

time, for some hypothetical future value of X may be under
consideration; in each case, the corresponding value of Y is
determined from the equation.
UNIT 3: Equations

25
Extending the example 3.1, let Y represent the value of the table,

S
Notes
the equation can be rewritten as:
___________________
7X + Y = 6,000
___________________
If X takes up the values 200, 300, 400 and 500 we can easily see

E
___________________
that Y will take the values 4,600, 3,900, 3,200, 2,500 respectively,
which are calculated by substituting the value of X in the above ___________________
equation. ___________________

Let us now plot this equation on the graph. On the graph the ___________________

UP
values of chair (X) are represented by rupees along the horizontal
___________________
line and value of table (Y) along the vertical line. Any convenient
scales may be chosen to represent the two variables; there is no ___________________
reason why the same distance as used to represent an increase of ` ___________________
100 on horizontal axis cannot be used to represent ` 1,000 on the
___________________
vertical axis.
E-
CC

Figure 3.1: Graph of 7x + y = 6,000

The point to be noted here is that both X and Y are dependent on

each other. If the value of X changes the value of Y also changes
and vice versa. This is because the net sum is a constant.

These kinds of equations are very easy to draw on a graph. You

substitute zero for one variable to get the corresponding value of
the other variable. In this example when we substitute zero for X
we get Y = 6000 and when we substitute zero for Y we get X = 6000
/ 7 = 857.14. When we want to plot this line on the graph we mark
6,000 on the vertical axis as this represents X = 0.
(c)

Similarly, we mark 857.14 on the horizontal axis as this represents

Y = 0. Connecting these two points by a straight line gives us the
line which can be used to find the value of one chair corresponding
to value of one table, satisfying the above equation.
 Quantitative Techniques for Management Applications

26
The word ‘linear’ is used to represent a function, which can be

S
Notes
represented by a straight line (and not any function which can be
___________________
represented by a line). Another point that must be remembered is
___________________ that although we are using letter Y to represent the dependent
___________________ variable, other letters can also be employed for denoting variables.

E
Like S can represent Sales, D can represent Distance and so on.
___________________
Once we have the equation plotted on a graph we can very easily
___________________
find out the value of one variable from the given value of another
___________________ variable. In the example 3.1, we were given the value of Y, i.e., the

UP
___________________ value of one table as ` 2,500. So we draw a straight line from

___________________

___________________

___________________
E-
Figure 3.2: Graph of 7x + y = 6,000

the vertical axis parallel to the horizontal axis at the value ` 2,500.
The point where this line meets the equation line is used to draw a
line parallel to the vertical line. Where this line meets the
horizontal axis gives us the value of one chair. So use of graphs
CC

makes it very easy for us to solve these linear equations.

What happens when we are not given the value of both the
variables? We cannot solve the equation if only a single equation is
given but only give a range of values, which these two variables
can take, all of which satisfy the equation in hand. But if we are
given two simultaneous equations then it makes it easy for us to
calculate the exact values of these two variables, which satisfy
both the equations simultaneously.
In general it is usually possible to solve a set of equations if the
number of variables is equal to number of equations, except in a
(c)

case where the equations overlap.

Example 3.2
A manufacturer of printed fabrics has three machines, that
prepare raw fabric and five machines that print on it. Two types of
printed fabrics are produced; type A requires 3 minutes per meter
UNIT 3: Equations

to prepare and 6 minutes per meter to print, while type B requires 27

S
11 and 17 minutes per meter respectively. How much of each type Notes
of fabric should be produced per hour in order to keep all the ___________________
machines fully occupied?
___________________
Solution:

E
___________________
The quantities to be produced per hour can be represented by X ___________________
meters of type A and Y meters of type B. Then the situation above
___________________
can be summarized in two simultaneous linear equations, one
equation for each machine ___________________

UP
3 X + 11 Y = 180 (1) ___________________

6 X + 17 Y = 300 (2) ___________________

___________________
The right-hand sides of these equations are obtained from the fact
that there are 180 machine-minutes available per hour for ___________________
preparing fabric (60 minutes x 3 machines) and 300 machine-
minutes for printing (60 minutes × 5 machines).

There are three ways of solving any pair of simultaneous linear

equations.
E-
Elimination Method:

It will be observed that 6X is exactly twice 3X and so the first

equation can be doubled to give:

6 X + 22 Y = 360

This is then subtracted from equation (2) to eliminate the terms

CC

involving X:

- 5 Y = -60

Y = 12

Substituting this value of Y in equation (1):

3 X + 132 = 180

X = 16

Substitution Method:

The second method of solution is by substitution. Equation (1) is

(c)

rearranged so that one of the unknowns is expressed in terms of

the other:

3 X = 180 - 11 Y

X = 180 - 11Y
3
 Quantitative Techniques for Management Applications

28
This formula for X is then substituted in equation (2):

S
Notes

___________________ 6 ( 180 -311Y ) + 17 Y = 300

___________________
360 – 22Y + 17Y = 300
___________________ 360 - 5 Y = 300

E
___________________ Y = 12
___________________ The value of X is then found using equation (1) and substituting
___________________ them in equation (2) can check both values.

UP
___________________ Cross Multiplication Method:
___________________ a1 x + b1 y + c1 = 0 ...(1)
___________________ a2 x + b2 y + c2 = 0 ....(2)
___________________
x y 1
a1 b1 c1 a1 b1
a2 b2 c2 a2 b2
x y 1
= =
E-
b1c 2 − b2 c1 a2 c1 − a1c 2 a1 b2 − a2 b1
x 1
∴ =
b1c 2 − b2 c1 a1 b2 − a2 b1

b1c 2 − b2 c1 b2 c1 − b1c2
∴ x= which is the same as x =
a1 b2 − a2 b1 a2b1 − a1b2
CC

y 1
and =
a2 c1 − a1c 2 a1 b2 − a2 b1

a2 c1 − a1c 2 a1c2 − a2 c1
∴ y= which is the same as x =
a1 b2 − a2 b1 a2b1 − a1b2
Note: While solving the system of linear equations, anyone of these
methods can be used.
Example 3.3
2 x + y = −1

x−y = 4
(c)

Solution:
(A) By substitution method:
2 x + y = −1 ...(1)

x-y=4 ...(2)
UNIT 3: Equations

29
(1) Substituting this in (2), we get

S
Notes
⇒ y = −2 x − 1.
___________________
x + 2 x + 1 = 4 ⇒ 3x = 3 ___________________

Substituting this value of x in (1), we get

E
___________________
3
∴x = i.e., x = 1 ___________________
3
___________________
2(1) + y = −1
___________________

UP
i.e. y = −3 ___________________

(B) By elimination method: ___________________

2 x + y = −1 ...(1) ___________________

___________________
x−y = 4 ...(2)

Adding (1) and (2), we get

3x = 3
E-
3
∴x = =1
3

i.e. x = 1

Multiply (2) by 2
(1) ⇒ 2 x + y = −1
CC

2 × (2) ⇒ 2 x − 2 y = 8

2 × (2) ⇒ 2 x − 2 y = 8

Subtracting, we get
∴ y = −3

(C) By cross multiplication method: The equations are:

2x + y + 1 = 0

x−y−4 =0

x y 1
(c)

2 1 1 2 1
1 −1 −4 1 −1

x y 1
= =
−4 + 1 1 + 8 −2 − 1
 Quantitative Techniques for Management Applications

30
x y 1

S
Notes i.e. = =
−3 9 −3
___________________
−3 9
i.e. ∴ x = ,y=
___________________ −3 −3
___________________

E
i.e. x = 1 , y = −3
___________________
The stepwise general procedure for solving these linear equations
___________________
is given below for your reference:
___________________
Box 3.1: Stepwise Procedure for Solving 2×2 simultaneous Equations

UP
___________________
1. Eliminate one of the variables using any or both of the properties
___________________
specified below-
___________________
(a) Any linear equation can be multiplied or divided on both sides
___________________
by any number without altering its truth or meaning.

(b) Any two linear equations can be added or subtracted (one from
the other) to give a third, equally valid, equation.

2. Solve the resulting simple equation (to yield the value of the other
E-
variable).

3. Substitute this value back into one of the original equations, say
equation 1 ( to yield the value of the first variable).

4. Check the solutions (by substituting both values into original

equation 2).
CC

Of course, graphical method can also be used to solve the 2x2

simultaneous equations. The first step is to draw the two lines
represented by the two given equations on the same graph and the
second step is to identify the X and Y values at the intersection of
the lines. These X and Y values are the required solution for the
pair of simultaneous equations.

Check Your Progress

Fill in the blanks:
1. Linear Equation may be defined as an equation where
the power of the variable(s) is ________, and no cross or
(c)

product terms are present.

2. The word ‘linear’ is used to represent a function, which
can be represented by a ___________.
UNIT 3: Equations

Graphical Solution 31

S
Notes
The situation above was summarized in two simultaneous linear
___________________
equations:
___________________
3 X + 11 Y = 180 (1)

E
___________________
6 X + 17 Y = 300 (2)
___________________
Plotting the two equations simultaneously on the graph we find ___________________
that they intersect at values Y=12 and X=16. This becomes the
___________________
solution to the problem.

UP
___________________
Similarly, 3 × 3 simultaneous equations can also be solved using
___________________
an extension of the same technique that we have used above. There
are other methods to solve these simultaneous equations which we ___________________

would discuss in subsequent discussion on matrices and ___________________

determinants and linear programming.
E-
CC

Figure 3.3: Graph of 3X +11Y=180 and 6X + 17Y = 300

The basic method for solving these 3 × 3 equations mathematically

is given below:
Box 3.2: Procedure for Solving 3 × 3 Simultaneous Equations

1. Using any two of the given equations, eliminate one of the variables
(using the equation-manipulating techniques previously described)
to obtain an equation in two variables.

2. Using another pair of equations, eliminate the same variable as in

(1), which will give a second equation in two variables.
(c)

3. Solve this 2 × 2 system of equations in the normal way.

4. Substitute into one of the three original equations to find the value
of the third variable.

5. Check the solutions by substituting the values of these variables in

the three equations.
 Quantitative Techniques for Management Applications

32
Example 3.4

S
Notes

___________________
A furniture manufacturer sends Company A a bill for ` 10, 700 to
cover 3 tables, 4 chairs and 3 stools. Company B is charged ` 14,
___________________
800 for 2 tables, 5 chairs and 7 stools. Company C is charged ` 15,
___________________ 100 for 5 tables, 9 chairs and 2 stools. What are the respective

E
___________________ prices for each of these items?

___________________ Solution
___________________ Representing the prices of one table, one chair and one stool by ` x,

UP
___________________ ` y and ` z respectively, the problem gives rise to three
simultaneous linear equations:
___________________
3 x + 4y + 3z = 107 (1)
___________________
2x + 5y + 7z = 148 (2)
___________________
5x + 9y + 2z = 151 (3)
These equations are still called ‘linear’ even though each could only
be represented by a plane in a three-dimensional model and not by
a straight line on a two-dimensional graph. The first step in their
E-
solution would be to multiply the first equation by 2 and the
second equation by 3 in order to eliminate x and then subtracting
first equation from the second one:

6x+ 8y + 6 = 214
6 x + 15y + 21z = 444
CC

7y + 15z = 230
The second and third equations are then multiplied by 5 and 2
respectively in order to obtain a second equation in which x has
been eliminated. The two equations involving only y and z are then
solved as in example 3.2, to give y = 5, z = 13. Substituting these
values in the first of the original equations gives x = 16.
Substituting them in the other two original equations can check all
three values.

Summary
(c)

While the functions tell us that a relationship exists, equations

give us the exact relationship between the variables. A linear
expression equated to zero is called a linear equation. Thus the
general form of a linear equation is ‘ax + by +c = 0’, where a, b and
c are real constants. An equation that can be written as ‘ax2 + bx +
C = 0’, is called a quadratic equation. An equation of degree n has
UNIT 3: Equations

exactly n roots. Hence, a linear equation has exactly one root and a 33

S
quadratic equation has exactly two roots. Notes

___________________
Lesson End Activity ___________________

Write the Procedure for solving 3 x 3 simultaneous equations

E
___________________
taking suitable example of the same. ___________________

___________________
Keywords
___________________

UP
Equation: A statement that two expressions (connected by the ___________________
sign=) are equal.
___________________
Linear Equation: A Linear expression equated to zero is called a ___________________
linear equation.
___________________

Questions for Discussion

1. Find two numbers whose difference is 8 and product is 20.
E-
2. Find two consecutive numbers such that 5 times the smaller
number is equal to 5 more than three times the greater
number.

3. Find two numbers whose sum is 32 and their product is 252.

4. 2 years ago, elder brother's age was 3 times the square of his
younger brother's age. After 3 years, elder brother's age will be
CC

4 times his younger brother's age. Find their present ages.

5. Find two consecutive integers such that 10 times the smaller

number is 5 times the bigger number.

6. When the solutions of quadratic equation are termed as:

(a) rational and equal,

(b) real and distinct and

(c) imaginary and distinct.

(c)

Further Readings

Books
R S Bhardwaj, Mathematics for Economics and Business, Excel
Books, New Delhi, 2005
 Quantitative Techniques for Management Applications

34
D C Sanchethi and V K Kapoor, Business Mathematics

S
Notes

___________________ Sivayya and Sathya Rao, An Introduction to Business Mathematics

___________________
Web Readings
___________________

E
www.textbooksonline.tn.nic.in
___________________
www.mathbusiness.com
___________________

___________________

UP
___________________

___________________

___________________

___________________
E-
CC
(c)
UNIT 4: Applications of Linear Equations and Functions in Business

Unit 4
35

S
Notes

Applications of Linear Equations

___________________

___________________

and Functions in Business

E
___________________

___________________

Objectives ___________________
After completion of this unit, the students will be aware of the following
___________________

UP
topics:
___________________
\ Applications examples
\ Supply and Demand Functions ___________________

\ Quadratic equations ___________________

___________________

Introduction
In this unit, we have focused on various application problems
related with the use of functions and equation. We have also
E-
focussed on supply and demand functions. We have tried to explain
the concept by providing number of examples.
Example 4.1
The total production costs of a packaging machinery manufacturer
are found to be an average of ` 60,000 per day. The cost accountant
finds that the fixed costs are ` 32,000 per day and the direct costs
CC

average ` 7000 per machine. Calculate the average number of

machines produced per day?
Solution:
This employs the accountants’ terms ‘fixed costs’ and ‘direct costs’
and uses the accountants model:
Total costs = fixed costs + (direct costs × quantity produced), i.e., T
= F + Dx
Let x represent the no. of machines sold, the above model would
look like:
(c)

` 60,000 = ` 32, 000 + (` 7,000 × x) (1)

Dividing the equation by ` 1,000 it reduces to
60 = 32 + 7x (2)
 Quantitative Techniques for Management Applications

36
In the above example, the model is very useful though

S
Notes
approximate, since the direct costs per machine will probably vary
___________________
quite widely. These types of models are used a great deal and are
___________________ regarded as absolute truth by top management. But these models
___________________ have their limitations as we will see.

E
___________________ Now if all the machines are sold at the same price, then the
revenue is a linear function of the quantity produced. Putting R for
___________________
revenue and p for price the function becomes
___________________
R = px

UP
___________________
In the case of machine manufacturer, let us assume that the
___________________
selling price is ` 18,000. By putting p = 18 (again assuming ` 1000
___________________ as the unit) a graph (Figure 4.1) can be drawn of
___________________ the revenue function. However, it is much more informative to
draw the line representing the cost function and revenue function
on the same graph, as shown in the graph.
Extending the lines beyond their range in which their practical
usefulness is proved is called extrapolation; it is
E-
a bad practice to extrapolate too far. The unreliable parts of the
lines on the graph are shown by broken lines and the meaningless
parts by dotted lines.
CC
(c)

Figure 4.1: Machine Manufacturers' Cost and Revenue Functions

UNIT 4: Applications of Linear Equations and Functions in Business

37
The difference between revenue and total production costs can be

S
Notes
described as gross profit G:
___________________
G=R – T
___________________
=px – (F+Dx) = 18x – (32 + 7x)

E
___________________
= 11x – 32
___________________
The break even point is when profit = 0, that is your revenue is ___________________
equal to your costs. Putting this in the above equation we get:
___________________

UP
11x – 32 = 0
___________________
11x = 32
___________________

32 ___________________
x= = 2.91(approx. )
11 ___________________
So your average production should be 2.91 machines for you to
cover all your costs but make no profits.
This breakeven level can also be found from the graph where your
E-
revenue and costs curves cross each other. Alternatively, you can
plot the equation (11x-32=0) and find the value of x where the line
meets the x-axis as at that point the value of the function would be
zero.
There are two possible ways in which you could have obtained the
information related to fixed and direct costs as the cost accountant
CC

found. Either you take all the accounting records and classify each
cost into the two headings, a tedious and time consuming process
which is prone to error because of limited accounting knowledge
and problems of classification.
Or a quicker and better method would be to record the actual total
costs at two different levels of production and then find linear cost
function which ‘fits’ these actual costs. For instance, the records
might show that the average total cost per day was ` 49,500 when
production averaged 2.5 machines per day and it rose to ` 63,500
when production rose to an average of 4.5 machines per day.
All that is necessary is to insert these two values of T and the
(c)

corresponding values of x into the linear cost function, defined

above.

T = F + Dx
This gives two equations, involving two unknowns F and D
 Quantitative Techniques for Management Applications

38
49.5 = F + 2.5D

S
Notes
Activity
What are demand and supply
___________________
63.5 = F + 4.5D
functions?
___________________ Solving these equations using the techniques already described, we
get D = 7 and F = 32.
___________________

E
Substituting these values in the above function we get T = 32 + 7x,
___________________
i.e., the same equation the cost accountant had with all his
___________________ information.
___________________
Check Your Progress

UP
___________________ Fill in the blanks:
___________________
1. If all the machines are sold at the same price, then the
___________________ revenue is a ___________ function of the quantity
___________________ produced.
2. Extending the linear beyond their range in which their
practical usefulness is proved is called ___________.

Supply and Demand Functions

E-
Supply and Demand Functions are an important field of study for
the economists. The amount of a particular product
which a firm is willing to supply at a specified price will depend on
the firms’ cost function and also on its marketing policy. The firm
may be concerned with maximizing profit, increase market share,
or just to keep the factory going in times of economic slow down.
CC

Once the firm decides what its policy is, the amount of products,
which can be supplied to the market, is clearly a function of the
price at which these products can be sold in the market. This forms
the supply function of the firm. If the quantities of products that
can be supplied by all the firms in this industry are totalled up for
each price level of the product, this gives us the total supply
function for the market as a whole.
As an example, let us assume that total supply of a particular type
of phones in the market is 29,000 pieces per month when the price
is ` 500 per piece. The same manufacturers are prepared to supply
a total of 52,000 pieces per month if the price is raised to
(c)

` 600 per piece. A further rise in the price per piece would justify
working overtime in the factory and also bring in foreign suppliers
who were earlier not interested in selling at low prices in the
market. It is found that a total of 75,000 pieces per month can be
supplied when the price is ` 700 per piece.
UNIT 4: Applications of Linear Equations and Functions in Business

39
These three individual points can be plotted on a graph (Figure

S
Notes
4.2) letting ` P represent the price per phone (in ` hundred) and X
the total quantity (in thousands of pieces per month) which would ___________________
be supplied at that price. Although, here P is the independent ___________________

E
___________________

___________________

___________________

___________________

UP
___________________

___________________

___________________

___________________
Figure 4.2: Demand & Supply Curves

variable and X is the dependent variable, it is customary for

economist to plot prices on the vertical axis and quantities on the
E-
horizontal axis and this practice would be followed here. In the
simplified example being considered, three points are found to lie
on a straight line and so it can be assumed that the supply
function for the market is approximately linear. The function is
then found to be:

X = 23P – 86
CC

For example, by substituting ` 500 as price (P = 5) we get X = 29

(i.e., 29,000 phones). This line represents the quantities which will
be produced at different prices provided all the quantity produced
can be sold. But to find out what can be sold in the market we need
look at the demand function of the market. The demand function
would indicate the total quantity that will be purchased at a
particular price and therefore, represents the total individual
demand functions of all the individual buyers.
Normally, large quantities would be bought when the price is lower
and as the price goes up the quantities purchased come down. In
this particular case it was found that only 24,000 telephone pieces
can be sold at ` 700 per piece but that the sales would increase to
(c)

35,000 and 46,000 pieces per month at the prices of ` 600 and ` 500
respectively. These three points can be plotted on the same graph
as supply curve so as to get the demand curve. The demand
function is then found to be:
X = 101 – 11P
 Quantitative Techniques for Management Applications

40
It would be wrong to assume that we can extend these lines on

S
Notes
Activity
either side for supply and demand functions. It would be absurd to
What are discontinuous
___________________
functions? assume that the demand is 2,000 pieces when the price is ` 900
___________________ and equally wrong to assume that demand is approximately 90,000
___________________ pieces when price is ` 100.

E
___________________ The reason for plotting supply and demand of the same graph is to
found out the point of market equilibrium, which is the point of
___________________
intersection of these two lines. It can also be find out using simple
___________________ equation solving techniques mentioned earlier, by finding the

UP
___________________ value of P which makes the value of X same for both demand and
___________________
supply functions. The prices and quantity at the point of market
equilibrium are known as equilibrium price and the equilibrium
___________________
quantity. Under the condition of free competition, the equilibrium
___________________ quantity will be the quantity actually produced and the
equilibrium price would be the price in the market.
Fitting demand and supply curves is much more tedious than
solving other business situations. It is much more difficult to
access how supply will respond to change in price than to access
E-
how the total production cost within a firm will vary with the
quantities produced. Demand is also complicated because of the
presence of substitute products in the market. These difficulties
explain why mathematical economists need a lot of training and
experience and why sometimes forecasted situations vastly differ
from the actual situations in the market.
CC

Check Your Progress

Fill in the blanks:
1. Fitting demand and supply curves is much ___________
than solving other business situations.
2. The reason for plotting supply and demand of the same
graph is to found out the point of _______________.

Irregular, Unequal and Discontinuous Functions

Sometimes it is assumed that ‘y is a function of x’ means that there
(c)

is a single formula connecting y with x. While it is easy to discuss

functions which are described by a single formula, the only correct
interpretation is that y is a function of x if the value of x
determines the value of y, irrespective of the fact, whether it is in
UNIT 4: Applications of Linear Equations and Functions in Business

steps, multiple formulae are required or there are constraints 41

S
attached. Notes

For example, since the price of a commodity determines ___________________

the quantity supplied and the quantity demanded, these quantities ___________________
are functions of the price, even in cases where the relationship is

E
___________________
so irregular that it can be described only by a list of prices with the
corresponding quantities. ___________________

Example 4.2 ___________________

___________________
A certain car is so expensive that wealthy people buy it for prestige

UP
reasons because it is expensive. The demand falls from 30 per ___________________
month when the price is ` 20,00,000 to 20 per month when the ___________________
price is `16,00,000 and then rises to 100 per month when the price
___________________
falls to ` 12,00,000.
___________________
E-
Figure 4.3: Demand for a Luxury Car
CC

Solution:
The methods discussed earlier can be used to find the equation for
the straight line which passes through the points (30, 20 lacs) and
(20, 16 lacs), and again for the straight line through the points (20,
16 lacs) and (100, 12 lacs). It is standard practice to denote points
in the graph in the form (x, y). In order to state the range of values
for which each of these equations is valid, it is necessary to use one
of the family of symbols known as inequalities. The most important
inequalities are

a > b means ‘a is greater than b’

(c)

a ≥ b means ‘a is greater than or equal to b’

a < b means ‘a is less than b’

a ≤ b means ‘a is less than or equal to b’

 Quantitative Techniques for Management Applications

42
Only the last one is needed for the present. The formulae for the

S
Notes
demand curve shown in the graph are:
___________________
x = (p/40000) – 20 when 16, 00,000 ≤ p ≤ 20, 00,000
___________________

___________________ x = 340 - (p/5000) when 12, 00,000 ≤ p ≤ 16, 00,000

E
___________________ The function is now defined for all values of p in the range 12 lacs
< p < 20 lacs. It is perfectly mathematically sound to leave the
___________________
function undefined outside this range if no information is
___________________ available, though a keen sales manager would like to see that

UP
___________________ whether it would be highly profitable to fix the price higher than `
20 lacs.
___________________

___________________
In this example, it is not correct to say that price p is a function of
quantity x. The value of x does not determine the value of p, since
___________________
p could be either of two values if the value of x is, for instance, 25.
For normal commodities the demand curves slope downwards
throughout their length and it is then correct to regard price and
quantity each as a function of the other.
E-
Some functions are represented by two or more lines which do not
meet each other. A good example is a schedule of postage rates
where the first slab is ` 2 up to 20 gms. and then Re 1 for every
additional 10 gms.
CC

Figure 4.4: Postal Rates Graph

Clearly, the postage rate is a function of the weight of the letter,

since the latter determines the former. Any function which consists
(c)

of two or more lines which do not meet is called a discontinuous

function. There is said to be a discontinuity at each of the values of
the independent variable at which there is a gap in the values of
the dependent variable.
UNIT 4: Applications of Linear Equations and Functions in Business

43
It must be noted that the graph includes the points (20, 30), (30,

S
Activity
Notes
40)…… but does not include the points (20, 40), (40, 80)……
When do we use quadratic
Sometimes appropriate marks are added to a graph to indicate ___________________
equations?
whether or not these boundary points belong to the function. ___________________

Discontinuities and inequalities introduce awkward complications,

E
___________________
and so mathematicians and scientists usually ignore them for as
___________________
long as they can. Unfortunately, they arise far too frequently in
___________________
managerial problems to be so lightly disregarded. Because of this
reason, we use limitations and constraints whenever we develop ___________________

UP
problems and analyze them. ___________________

Check Your Progress ___________________

Fill in the blanks: ___________________

___________________
1. Any function which consists of two or more lines which
do not meet is called a ___________ function.
2. There is said to be a discontinuity at each of the values
of the ___________ variable at which there is a gap in the
E-
values of the ___________ variable.

Quadratic Equations
We saw that it is usually possible to sell larger quantities of a
commodity where the price is lower. For a monopolist, the demand
curve for the market is the price curve to be used in calculating the
CC

revenue of the firm. Where there is no monopoly, the amount a

manufacturer can sell is still a function for the price at which he
offers his goods, although in this case the price curve will not be
the same as the demand curve for the market as a whole.
Let us consider an example.
Example 4.3 (Extending example 4.1)
The same machine manufacturer finds that he could sell an
average of four machines per day at a price of ` 18,000 per
machine. Stepping up his production to an average of 4½ machines
per day, he finds that he has to reduce the price to ` 17,500 per
machine in order to sell all that he produces. Find the profit
(c)

function.
Solution:
Now this problem abandons the unrealistic assumption of
traditional cost accounting that the price is a constant and the
 Quantitative Techniques for Management Applications

44 revenue function therefore linear. Putting the machines sold per

S
Notes day, x, as a linear function of the price (in units of ` 1,000), p:
___________________
x = ap + b
___________________
The method learned earlier makes it easy to find out a and b.
___________________ Substituting their values we find that the equation reduces to:

E
___________________ x= 22 – p
___________________
The revenue R is the price multiplied by the number of machines
___________________ sold:

UP
___________________ Revenue R = Price p x Quantity x
___________________ R = px
___________________
In order to find the breakeven point, it is simplest to express p as a
___________________ function of x; R then becomes a quadratic function of x:
p = 22 – x from the equation x = 22 - p above
Substituting this value of p in R = px we get
R = px = (22 – x) x = 22x – x2
E-
W
Assuming that the linear cost function to be 32 + 7x, as found m
earlier, the gross profit G becomes a quadratic function of x:
G=R–T
= (22x – x2) – (32 + 7x) = -x2 + 15x – 32
This is the profit function for this manufacturer.
CC

This quadratic function more closely approximates the real life

situation. Now the question comes, how do we solve these
quadratic functions/equations. There are three basic methods:

1. Factorization

2. Using Graphs

3. Using Formula

Solution of Quadratic Equations by Factorization

If the quadratic equation can be expressed as a product of two
(c)

linear expressions (known as factors) it can be solved using

factorization. This is possible only if the solution to the equation is
an ordinary number or a fraction. For example:

2x2 – 17x + 21 = (x – 7) (2x – 3) = 0

UNIT 4: Applications of Linear Equations and Functions in Business

45
The identity sign (=) is used as a reminder that the two sides are

S
Notes
equal for all values of x, which can be confirmed by multiplying out
the right-hand side. ___________________

If the product of any two expressions is zero, then at least one of ___________________

these expressions must be zero. So recognizing the factors

E
___________________
immediately leads to the solution of the equation:
___________________
2x – 17x + 21 = 0
2
___________________
(x – 7) (2x – 3) = 0 ___________________

UP
Either x – 7 = 0 or 2x – 3 = 0 ___________________

x = 7 or x = 32 ___________________

___________________
The main difficulty in finding the solution of a quadratic equation
___________________
by factorization lies in finding the factors. It is possible to do this
by a routine procedure which will either find the factors
systematically or prove that none exists.

Procedure for Finding Factors

E-
Let the factors be (px + q) and (rx + s), where p, q, r and s are
positive or negative integers and the product of the two factors is
ax2 + bx + c. Multiplying the factors we get:
prx2 + psx + qrx + qs = ax2 + bx + c
The coefficients must be the same on both sides as this must be
CC

true for all values of x. This gives equations relating the unknown
quantities to the coefficients in the expression to be factorized: pr =
a; ps + qr = b; qs = c. It implies that the product of ps and qr is ac
and so the first task is to find these two numbers whose sum and
product are known.
Example 4.4

Find the factors of 10.8x2 + 93x + 140

Solution:
The first stage is to take out the fractional factor, resulting in the
expression 0.2 (54x2 + 465x + 700). It is then necessary to look for
(c)

two numbers whose sum is 465 and whose product is 54 × 700 =

37,800.
The simplest method is to start with any two numbers whose sum
is 465, such as 80 and 385. If the product is too small, a suitable
amount is added to the smaller one of the two and an equal
 Quantitative Techniques for Management Applications

46 amount is subtracted from the larger one. A little exercise will

S
Notes show that the numbers are 105 × 360 = 37,800.
___________________
Putting ps = 105 and using the fact that pr = 54, the highest
___________________ common factor 3 is then equated to p. Calculations would show
___________________ that s = 35, r = 18 and q = 20 and the factors of the above equation

E
are:
___________________
0.2 (3x + 20) (18x + 35)
___________________
Example 4.5
___________________

UP
___________________ Find the factors of 11x2 + 56x + 21.

___________________ Solution:

___________________ Two numbers are to be found whose sum is 56 and the product is
11 × 21 = 231. Following the above procedure, it will be found that
___________________
4 × 52 is too small and 5 × 51 is too large. It can immediately be
concluded that there are no integral factors.
Example 4.6
Find the factors of 12x2 + 39x – 105.
E-
Solution:
Here two numbers have to be found whose sum is 39 and the
product is 12 × (-105) = -1260.
It requires thinking to find that the one which is larger
numerically (that is, disregarding any negative sign) must be
CC

positive and the other negative. Trying 70 × -31 = -2170 is too big;
60 × –21 proves correct. It is pointless to try any pair of numbers
which does not include a multiple of 5 because the product is
required to be a multiple of 5.
Depending on which number is chosen to represent ps, the factors
are either (12x – 21) (x + 5) or (3x + 15) (4x – 7), both of which can
be reduced to 3(4x – 7) (x + 5). It would have been better to take
out the factor 3 right at the beginning if it had been recognized as
a common factor of all the coefficients.

Solution of Quadratic Equations Using Graphs

(c)

One way of finding the solution of any equation which is in the

form or can be rearranged in the form is:
f(x) = 0
UNIT 4: Applications of Linear Equations and Functions in Business

47
is to draw the graph of the function:

S
Notes
Y = f(x)
___________________
If this graph cuts the x-axis at any point, the value of x at that ___________________
point is the solution of the original equation, since it is the value of

E
___________________
x at which y = 0. This was the method discussed earlier also for
___________________
linear equations and can be applied to quadratic equation ax2 + bx
+ c as well. Here a, b and c can be positive or negative and may ___________________

involve fractions or decimals. It is also possible for b or c to be zero, ___________________

UP
but if a was zero, the function would become a linear function. ___________________

Example 4.7 (example 4.3 extended) ___________________

Plot the function G = –x2 + 15x – 32 on a graph. ___________________

___________________
Solution:

To draw the graph of the function:

G = –x2 + 15x – 32
E-
It is necessary to choose a range of values of x and calculate the
corresponding values of G. In this example it is enough to consider
values of x between 0 and 14:
x : 0 2 4 6 8 10 12 14
G : -32 –6 12 22 24 18 4 –18

It can be seen from the graph below that G is zero when x is about
CC

2.6 or about 12.4. These two values are said to be the roots of the
equation –x2 + 15x – 32 = 0.

This means that the manufacturer will make a profit between 2.6
to 12.4 machines and would make the maximum amount of profit
when he makes 8 machines.

Every quadratic function in which the coefficient of x2 is negative

gives a graph of shape shown above, which is termed a parabola. If
the highest point is above the x-axis, the corresponding equation
has two roots, but if the highest point of the function is below the
(c)

x-axis there is no solution for the corresponding quadratic

equation.
 Quantitative Techniques for Management Applications

48

S
Notes

___________________

___________________

___________________

E
___________________

___________________

___________________

UP
___________________

___________________

___________________

___________________ Figure 4.5: Graph of -x2 + 15x -32=0

Solution of Quadratic Equation by Formula

It is necessary to have a method of solving an equation in which
E-
the quadratic expression is difficult to factorize such as examples
4.3 and 4.5.. The method used is equally applicable to equations
where factorization is simple.

The derivation of the formula is of interest only to mathematicians

the solution is given below for a quadratic equation ax2 + bx + c =
0.
CC

2
-b ± b - 4ac
X=
2a
This formula can be applied to any quadratic equation irrespective
of the fact whether the coefficients are positive or negative.
Example 4.7 (Cont....)
Taking the equation given in the example 4.7, here
a = -1, b = 15 and c = -32. The solution by formula is:

- 15 ± (15) 2 - 4 (-1) (-32)

x=
(c)

2(-1)

=
-15 ± a225 - 128f = -15 ± 9.85
-2 -2

= 12.425 or 2.575.
UNIT 4: Applications of Linear Equations and Functions in Business

This is the exact solution whereas from the graph we got an 49

S
approximate solution. The sum of the roots is always equal to -b/a Notes
Activity
and their product is always equal to c/a. In this example the sum Discuss fitting a quadratic cost
___________________
curve.
15.05 and the product 32.62 can be checked with the original
___________________
equation. This check makes it unnecessary for you to check the
roots separately by substitution in the equation.

E
___________________

___________________
Check Your Progress
___________________
Fill in the blanks:
___________________
1. If the quadratic equation can be expressed as a product

UP
of two linear expressions then the two expression are ___________________

called ______________. ___________________

2. Every quadratic equation in which are coefficient of x2 is ___________________

negative gives a graph of _________ shape. ___________________

Fitting a Quadratic Cost Curve

A linear function is never an ideal model of production costs. It is
usually possible to obtain a much better model by fitting a
E-
quadratic curve as we did in the example 4.7 above.
T = ax2 + bx + c
The values of a, b and c will be positive. The term x2 implies that
costs increase more steeply as the production goes up. If there was
no such effect, it would never pay to enlarge a factory.
It must not be assumed that a quadratic curve will be a perfect
CC

model of production costs. There is nothing magical about the x2

term. A quadratic function will always be as good as a linear
function and nearly always be a better deal.
Example 4.8 (Extension of example 4.1)
The machinery manufacturer, discussed earlier, found that the
total production costs averaged ` 60,000 per day when an average
of 4 machines per day are produced. An accurate assessment of
costs when the average production is 3½ machines per day and
again at 4½ machines per day gives figures of ` 56,600 and `
63,600 respectively. Fit a quadratic cost curve.
(c)

Solution

Just as fitting a linear curve to two known points was shown

earlier to give two simultaneous linear equations, so fitting a
quadratic curve to three known points give three simultaneous
 Quantitative Techniques for Management Applications

50
linear equations. From the given information, again using units of

S
Notes
Activity ` 1000, the equations are:
What is the way for fitting a
___________________
quadratic cost curve? a (3½)2 + b(3½) + c = 56.6 (1)
___________________

___________________ a (4)2 + b(4) + c = 60.0 (2)

E
___________________ a (4½)2 + b(4½) + c = 63.6 (3)
___________________
This set of equations can be solved very easily by elimination,
___________________ eliminating first c and then b to give a = 0.4, b = 3.8 and c = 38.4.

UP
___________________
The cost function is therefore:
___________________
T = 0.4x2 + 3.8x + 38.4 (4)
___________________
When the quadratic revenue curve found in example 4.4 is applied,
___________________
the gross profit function is found to be:

G = -1.4x2 +18.2x – 38.4 (5)

The breakeven point is then approx. 2.65, compared with 2.58

E-
when a linear cost curve is assumed.

More advanced techniques based on more complicated models are

available in managerial problems, but the practical benefits of
increased accuracy will be negligible in most cases.

Summary
CC

There are two possible ways in which you could have obtained the
information related to fixed and direct costs as the cost accountant
found. Either you take all the accounting records and classify each
cost into the two headings, a tedious and time consuming process
which is prone to error because of limited accounting knowledge
and problems of classification.

Lesson End Activity

Solve the following equations by formula method and by
factorisation method:
(c)

(a) x 2 − 2 x + 1 = 0

(b) x 2 − 5x + 6 = 0
1 1 1 1
(c) − = −
x+1 x−1 x−2 x+2
UNIT 4: Applications of Linear Equations and Functions in Business

Keywords 51

S
Notes
Quadratic Equation: An equation of the form, where a 0, b, c are
___________________
constant, is called a quadratic equation.
___________________
Root: A value of the variable which satisfies the given equation is

E
called a solution or root. ___________________

___________________
Questions for Discussion ___________________

1. Discuss any two methods of finding the solutions of a system ___________________

UP
of linear equations. ___________________
2. An investor wants to invest ` 15,000 in two types of bonds. He ___________________
earns 12% in first type and 15% in the second. Find his
___________________
investment in each of his total earning is ` 1950.
___________________
3. M/s Kalyani Forge pays its workers ` 70 for an 8-hour shift. In
addition each worker is paid ` 10 for every one hour of
overtime. However, overtime cannot exceed 4 hours per day.

(a) Cite the total wage paid to the worker as a function of

E-
overtime.
(b) Draw the graph of this function.
4. Ash Lubes sells X units of Supreme Lubes each day at the rate
of ` 50 per unit of 100 gm. The cost of manufacturing and
selling these units is ` 35 per unit plus a fixed daily overhead
cost of ` 10,000. Determine the profit function. How would you
interpret the situation if the company manufactures and sells
CC

400 units of the lubes a day?

5. Parker India Ltd., manufacturers of quality stationary items,
has introduced a new variety of pen in the market. The
market supply function of the pen is represented by the
function Q = 160 × 8P, where Q denotes the quantity supplies
and P denotes the market price per unit. It costs ` 4 to
produce a pen. If the total profit required is ` 500 what should
be the market price per unit?
6. Ishaan Petrochemicals has introduced in the market its latest
lube. The marketing manager has worked out that the
demand function of this product, which can be expressed as:
(c)

Q = 30 – 4P,
Where, Q is the quantity and P is the per kilogram price.
(a) Write the total revenue as a function of price.
(b) Draw the graph of this function.
 Quantitative Techniques for Management Applications

52
7. The monthly supply of 2T Oil in Delhi is estimated to be

S
Notes
95,000 tons when the price is ` 13,000 per ton and 1,10,000
___________________
tons when the price is ` 16,000 per ton. The monthly demand
___________________ is estimated to be 109,000 tons at ` 13,000 per ton and 99,000
___________________
tons at ` 16, 000 per ton. Assuming that the supply and

E
demand functions are both linear find these functions and
___________________ hence determine the equilibrium price and quantity.
___________________
8. A manufacturer of petrochemicals finds that his total
___________________ production cost is ` 1,20,66,000 per week when he is producing

UP
___________________
1240 tons per week. The fixed costs are ` 67, 34,000 per week,
and the selling price is ` 11, 700 per ton. Find (a) the weekly
___________________ revenue, (b) the weekly gross profit, and (c) the weekly
___________________ production and total production cost at the break-even point.
___________________ 9. Yarn is prepared from cotton by being passed successively
through slubbing, roving and spinning frames. Each 20 kg of
yarn A requires 6 minutes in a slubbing frame, 18 minutes in a
roving frame, and 106 minutes in a spinning frame. For yarn B
the times are 7, 27, and 150 minutes respectively, and for yarn C
E-
they are 8, 30 and 181 minutes respectively. If the plant consists
of eight slubbing frames, 28 roving frames and 162 spinning
frames, how much of each type of yarn should be produced per
hour in order to keep all the machines fully occupied?

10. A rubber glove manufacturer finds that he can sell 1, 38,000

gloves per week pack sizes at ` 190 per pack. He increases the
price to ` 200 per pack and finds that he can sell only 1,
CC

28,000 packs `, per week. Assuming that price curve is linear,

find (a) the price, (b) the weekly revenue of the number of
packs sold per week, (c) find the prices and quantities for
which the weekly revenue will be ` 60,990 per week.

Further Readings
Books
R S Bhardwaj, Mathematics for Economics and Business, Excel
Books, New Delhi, 2005
D C Sanchethi and V K Kapoor, Business Mathematics
(c)

Sivayya and Sathya Rao, An Introduction to Business Mathematics

Web Readings
www.textbooksonline.tn.nic.in
www.mathbusiness.com
UNIT 5: Case Study

Unit 5
53

S
Notes

Case Studies
___________________

___________________

E
___________________
Objectives
___________________
After analyzing these cases, the student will have an appreciation of the
concept of topics studied in this Block. ___________________

___________________

UP
___________________
Case Study 1: Gwalior Drums Ltd.
___________________
Gwalior Drums Ltd. is a medium scale company, engaged in the
manufacture of drums of different qualities and sizes. It has a ___________________
fixed cost of `10,000,000. The average cost of manufacturing a
___________________
drum costs company ` 60 which the company sells at ` 100
assuming that every drum produced is sold off, find a formula for
profit for the company.

Question
E-
Find the minimum number of drums that the company should
produce and sell to meet exactly the cost.
Source: MBCQ-721D_Quantitative Techniques for Management Applications_CCE_UPES
CC
(c)
 Quantitative Techniques for Management Applications

54
Case Study 2: Cost & Revenue Calculation

S
Notes
A switch manufacturer finds that his total monthly production
___________________ costs are ` 10,600 when production is 16,000 units per month, `
___________________
17,800 when it is 26,000 units and ` 27,000 when the production
is 36,000 per month. He can sell 16,000 units per month at ` 104
___________________ each, but has to reduce the price to ` 94 each in order to sell

E
___________________ 26,000 pieces. He can sell 36,000 pieces only at ` 80.

___________________ Question

___________________ Assuming that both cost curve and price curve are quadratic, find

UP
___________________ (a) the monthly total cost,

___________________ (b) the price,

___________________ (c) the monthly revenue, and

___________________ (d) the monthly gross profit as functions of the quantity sold.

Find also

(e) the quantity sold,

(f) the price and

E-
(g) the monthly revenue at the breakeven point and confirm that
the monthly total cost is then equal to the monthly revenue.
Source: MBCQ-721D_Quantitative Techniques for Management Applications_CCE_UPES
CC
(c)
UNIT 6: Matrices

55

S
Notes

___________________

___________________

E
___________________

___________________

___________________

___________________

UP
___________________

___________________

___________________

___________________

BLOCK-II
E-
CC
(c)
Detailed Contents Quantitative Techniques for Management Applications

56

S
Notes
UNIT 6: MATRICES
___________________ z Objective and Subjective Probabilities
z Introduction z Basic Statement of Probability
___________________
z Vectors z Mutually Exclusive Events
___________________

E
z Multiplication of Vectors z Dependent and Independent Events
___________________
z Matrices z Decision Trees
___________________
z Properties of Matrix Multiplication and Addition z Revision of Probabilities
z
___________________
Use of Matrices for Production Planning Combinations

UP
z
___________________
UNIT 7: DETERMINANTS UNIT 9: RANDOM VARIABLES AND
___________________
z Introduction PROBABILITY DISTRIBUTIONS
___________________ z Introduction
z Solving Linear Equation by the Use of Determinants
___________________ z Random Variables
z The Best Method of Solving Linear Equations
z The Bernoulli Process
z Higher Order Determinants
z The Binomial Theorem
z Main Properties of Determinants
z Probability Distributions
z Applications in Management
E-
z Poisson Distribution
UNIT 8: PROBABILITY z The Normal Probability Distribution
z Introduction
z Concept of Probability UNIT 10: CASE STUDIES
CC
(c)
UNIT 6: Matrices

Unit 6
57

S
Activity
Notes
Define the term vector

Matrices
___________________
algebra.
___________________

E
___________________
Objectives
___________________
After completion of this unit, the students will be aware of the following
topics: ___________________

\ Vectors ___________________

UP
\ Multiplication of vectors ___________________
\ Matrices
___________________
\ Types of matrices
___________________
\ Use of Matrices for Production Planning
___________________

Introduction
If the economic model is extended to include several commodities,
it will have several equations in several variables. If all the
E-
equations of the model are linear, then matrix algebra provides an
efficient method of their solution than the traditional method of
elimination of variables. Matrix algebra also has the advantage of
presenting a system of several equations in a compact form.
It may be pointed out here that a given system of equations may or
may not be consistent. The method of testing the consistency of a
CC

system is also provided by the matrix algebra.

Since matrix algebra deals with a system of linear equations only,
it is also referred to as Linear Algebra.
Matrices form one of the most powerful management tool of
modern mathematics. They have innumerable applications in the
analysis of material and machine requirements and the solution of
problems in planning and organization. An understanding of
matrices is also essential for most branches of advanced
mathematics and statistics. As vectors lie at the base of matrices,
let us start by understanding them first.
(c)

Vectors
The use of vectors can be illustrated by a very simple example. A
small firm uses sheeting fabric to manufacture white sheets and
pillowcases for hospitals and hotels, which are sold by the dozen.
 Quantitative Techniques for Management Applications

58 Orders received in the office are passed by telephone to the

S
Notes
Activity packing department, who is interested only in the quantity to be
Define the term vector
___________________ packed in each parcel.
___________________ Typical orders would be ‘4 dozen sheets and 2 dozen pillowcases’,
___________________ ‘18 dozen sheets and 6 dozen pillowcases’, ‘12 dozen sheets’, ‘6

E
dozen pillowcases’ and so on. It would not be long before speaker
___________________
and hearer agree to save a lot of time and breath by giving simply
___________________ a pair of numbers for each order:
___________________
[4 2] [18 6] [12 0] [0 6]

UP
___________________
Here the first number stands for dozens of sheets and the second
___________________ number stands for pillowcases. The four brackets denotes four
___________________ different orders. As long as the zero is inserted when necessary,
there can be no confusion as to the meanings of these figures. As
___________________
the orders are packed, the quantities can be added up.
These pairs of numbers are examples of vectors. A vector is any
row or column of figures in a specified sequence. The fact that
[12 0] is an order for 12 dozen sheets while [0 12] would be an
E-
order for 12 dozen pillowcases indicates that the numbers acquire
meaning from their positions in the sequence. A vector is normally
printed between square or curved brackets or between a pair of
double vertical lines.
The sum of the four orders is an example of vector addition. Two
vectors are added together by adding the first number in the first
CC

vector to the first number in the second vector, the second number
in the first vector to the second number in the second vector and so
on. Each number is called an element of the vector. Vectors can
have more than two elements, but two vectors can only be added
together if both have the same number of elements. Clearly the
sum of the above four orders is [34 14], i.e. 34 dozen sheets and 14
dozen pillowcases.
If the firm started to sell blankets also, a new convention would be
needed by which [4 2 3] means 4 dozen sheets, 2 dozen pillowcases
and 3 dozen blankets. The convention would have to be adopted
completely for all orders, inserting 0 whenever an order did not
(c)

include any blankets. The total quantities ordered would be given

by the sum of these three-element vectors, which would itself be a
three-element vector. Vectors are, thus, an ordered arrangement of
numbers – it can be in a row or a column.
UNIT 6: Matrices

Multiplication of Vectors 59

S
Notes
Activity
If the customer, responsible for the order [4 2], asked for it to be Why it is meaningless to
___________________
doubled, this would be interpreted as [8 4]. If he asked for it to be multiply two vectors?
tripled, it would become [12 6]. This is the rule for multiplying a ___________________

vector by an ordinary number, which is called a scalar to

E
___________________
distinguish it from a vector. Hence, the result of multiplying a ___________________
vector [a b c] by a scalar k is the vector [ka kb kc].
___________________
A vector may also multiply a vector. But it would be meaningless ___________________

UP
to multiply together two vectors, both of which represent orders for
___________________
goods. The definition of vector multiplication will be seen to make
___________________
sense only when it is applied in a sensible situation.
___________________
When sheets and pillowcases have been ordered and packed, the
___________________
next stage is to invoice them. If the prices are ` 1,800 for a dozen
sheets and ` 700 for a dozen pillowcases, then the amount due for
the order [4 2] will be:

(4 × 1,800) + (2×700) = ` 8,600

E-
This suggests a use for the multiplication of vectors. The prices can
be represented by a new vector, which, because it is a different
kind of vector, will be written as a column:
LM1800OP
N700 Q
CC

Then multiplying an order vector by this price vector can be

defined as multiplying the first element of the order vector by
1,800 and the second element of the order vector by 700 and adding
the results together:

42
LM1,800OP = 8,600 18 6
LM1,800OP = 36,600 12 0
LM1,800OP = 21,600
N700 Q N700 Q N700 Q
0 6 M
L1,800OP = 4,200 34 14 M
L1,800OP = 71,000
N700 Q N700 Q
It is obvious that (34 x 1800) is the total value of all the sheets in
(c)

the preceding four order vectors and (14 × 700) is the total value of
all the pillowcases, so that the sum of these products, 71000, must
be the sum of all the separate orders:

8,600 + 36,600 + 21,600 + 4,200 = ` 71,000

 Quantitative Techniques for Management Applications

60
Two vectors can be multiplied together only if both have the same

S
Notes
Activity
number of elements. Multiplication of a row vector by a column
Define the term matrices.
___________________
vector, which always results in a scalar, is called the scalar
___________________ multiplication of vectors.
___________________ Note that the product has meaning in this case only because the

E
___________________ first element in both the order vector and the price vector
represents dozens of sheets and the second element in each type of
___________________
vector represents dozens of pillowcases. Yet they are also
___________________ distinctive; one vector gives the number of units ordered and the

UP
other vectors the price per unit, so that the product gives the value
___________________
of the units ordered. It would be just as meaningless to add an
___________________ order vector to a price vector as it would be to multiply an order
___________________ vector by another order vector. Vectors have, in fact, been implicit
in some of the earlier examples in this book even though they were
___________________
not made explicit.

Check Your Progress

Fill in the blanks:
E-
1. A vector __________ multiply a vector.
2. It would be __________to multiply together two vectors,
both of which represent orders for goods.

Now let us turn our attention to matrices.

Matrices
CC

A matrix is a rectangular or square array of numbers arranged

into rows and columns, where the numbers acquire meaning from
their position in the array. This means that vectors we discussed
earlier are just simple example of matrices. Let us take up an
example.

Example 6.1

Let us assume that the manufacturer of sheets and pillowcases

discussed earlier has three types of machines. There is one
machine for cutting the fabric, three machines for sewing and one
for folding. The manufacturing times in minutes per dozen are:
(c)

Cutting Sewing Folding

Sheets 8 38 14

Pillowcases 6 32 4
UNIT 6: Matrices

To form these facts into a matrix, it is only necessary to arrange 61

S
the numbers between brackets or double vertical lines: Notes

LM8 38 14OP ___________________

N6 32 4 Q ___________________

E
___________________
Single vertical lines will not do, as these are used to represent a
determinant. Even when it has the same number of rows as ___________________
columns, a matrix is not at all the same thing as a determinant. ___________________
There is no way of expanding or evaluating a matrix, since each
___________________
element has its own distinctive meaning. The production time for

UP
an order [4 2] can be calculated by multiplying the order with the ___________________
manufacturing time. ___________________

For this purpose, each column of the matrix will be treated as a ___________________
column vector and the scalar multiplication of the order vector by
___________________
these column vectors would give the three results.

(4 × 8) + (2 × 6) = 44 minutes cutting

(4 × 38) + (2 × 32) = 216 minutes sewing

E-
(4 × 14) + (2 × 4) = 64 minutes folding

However, it is not necessary to separate out the column vectors; a

convention of matrix multiplication is adopted which gives the
same result:

LM8 38 14OP
N6 32 4 Q
CC

[4 2] = [44 216 64]

The production times now appear as the elements in a new row

vector. Vectors are really simple examples of matrices. In general,
a matrix has m rows and n columns. If m = 1, then the matrix is a
row vector with n elements. If n = 1, then the matrix is a column
vector with m elements. All general statements that we made or
will make about matrices will also apply to vectors as well.
Two matrices can be multiplied together only if the number of
columns in the first matrix is equal to the number of rows in the
second matrix. The first row of the first matrix is then multiplied
(c)

by the first column of the second matrix, following the rules for the
scalar multiplication of vectors and the result becomes the first
element in the first row of the matrix forming the answer. In
general, the product of the ith row of the first matrix and the jth
column of the second matrix becomes the element in the ith row and
the jth column of the matrix forming the answer.
 Quantitative Techniques for Management Applications

62
Therefore, it follows that a matrix with m rows and n columns can

S
Notes
be multiplied by a matrix with p rows and q columns only when n
___________________
is equal to p. The product is then a matrix with m rows and q
___________________ columns. In the above multiplication of a row vector by the
___________________ manufacturing time matrix, m = 1, n = 2, p = 2, and q = 3. Let us

E
now apply the rules of matrix multiplication.
___________________
The original four order vectors can be formed into a matrix in
___________________
which each of the four rows represents a different order. The
___________________ manufacturing time matrix can then multiply this and the answer

UP
___________________ is a matrix in which the columns represent the three types of
___________________
machine and the rows represent the time taken to process each of
the four orders.
___________________
LM4 2 OP LM44 216 64 OP
___________________
MM18 6 PP LM8 38 14OP = MM180 876 276 PP
MN012 60PQ N6 32 4 Q MN36 96 456 168
192 24 Q
P
Types of Matrices
E-
1. Rectangular Matrix: A matrix consisting of m rows and n
columns, where m ≠ n, is called a rectangular matrix. For
⎡a a12 a13 ⎤
example, A = ⎢ 11 is a rectangular matrix of order
⎣ a21 a22 a23 ⎥⎦
⎡ b11 b12 ⎤
2 × 3. Similarly B = ⎢b21 b22 ⎥ is a matrix of order 3 × 2.
⎢ ⎥
CC

⎢⎣b31 b32 ⎥⎦

2. Square Matrix: If the number of rows of a matrix is equal to

its number of columns, the matrix is said to be a square
⎡ a11 a12 a13 ⎤
⎢ a23 ⎥⎥ is a square matrix.
matrix. For example: A = ⎢ a21 a22
⎢⎣ a31 a32 a33 ⎥⎦

3. Row matrix: A matrix having only one row is called a row

matrix (or row vector). For example, [ 4 1 2 7 ] is a 1× 4
row matrix.
(c)

4. Column Matrix: A matrix having only one column is called a

⎡2⎤
column matrix (or column vector). For example, ⎢1 ⎥ is a 3 × 1
⎢ ⎥
⎢⎣4 ⎥⎦
column vector.
UNIT 6: Matrices

63
Diagonal Matrix: A square matrix A = ⎡⎣ aij ⎤⎦

S
5. is said to be Notes
Activity
n× n

diagonal matrix if aij = 0 for i ≠ j . For example, the matrix D What is scalar matrix?
___________________
given below is a 3 × 3 diagonal matrix. ___________________

⎡d1 0 0⎤

E
___________________
D = ⎢⎢ 0 d 2 0 ⎥⎥ ___________________
⎢⎣ 0 0 d 3 ⎥⎦ ___________________

We note that the elements of matrix A, for i = j, are called the ___________________

UP
diagonal elements and the line along which they lie is called ___________________
principal diagonal.
___________________
6. Scalar Matrix: A diagonal matrix in which all the diagonal ___________________
elements are equal, is called scalar matrix. The matrix
___________________
⎡2 0 0⎤
T = ⎢⎢ 0 2 0 ⎥⎥ is a 3 × 3 scalar matrix.
⎢⎣ 0 0 2 ⎥⎦

7. Identity (or Unit) Matrix: A square matrix with each of its

E-
diagonal elements equal to unity and all non-diagonal
elements equal to zero, is called an identity matrix. The
⎡1 0 0 ⎤
matrix I = ⎢0 1 0 ⎥ is a 3 × 3 identity matrix. We note that an
⎢ ⎥
⎢⎣ 0 0 1 ⎥⎦
identity matrix is a particular case of a scalar matrix where
CC

each of its diagonal elements is equal to unity.

8. Null (or Zero) Matrix: A matrix (square or rectangular)

having all its elements equal to zero, is called a null matrix
⎡0 0 0⎤
e.g. O = ⎢ ⎥ is a 2 × 3 null matrix.
⎣0 0 0⎦
9. Triangular Matrix: A triangular matrix can be: (a) Upper
triangular or (b) Lower triangular matrix.

(a) A square matrix A = [ aij ] is said to be upper triangular if

⎡ a11 a12 a13 ⎤
aij = 0 for i > j . The matrix A = ⎢⎢ 0 a23 ⎥⎥ is upper
(c)

a22
⎢⎣ 0 0 a33 ⎥⎦
triangular.
 Quantitative Techniques for Management Applications

64
(b) A square matrix A = [ aij ] is said to be lower triangular if

S
Notes

___________________ ⎡ a11 0 0⎤
___________________
aij = 0 for i < j . The matrix A = ⎢⎢ a21 a22 0 ⎥⎥ is lower
⎢⎣ a31 a32 a33 ⎥⎦
___________________

E
triangle.
___________________
Remarks:
___________________

___________________ 1. A diagonal matrix is both an upper and lower triangular.

UP
___________________ 2. If the diagonal elements of a triangular matrix are all zeros, it
___________________
is said to be strictly triangular.

___________________ Check Your Progress

___________________ Fill in the blanks:
1. Two matrices can be multiplied together only if the
number of columns in the first matrix is ________ to the
number of rows in the second matrix.
E-
2. A ____________ matrix is both an upper and lower
triangular.

Properties of Matrix Multiplication and Addition

It is often useful to represent a matrix by a single symbol. A capital
letter is usually printed in bold type to emphasize that it is a
CC

matrix. Putting the matrix of the four orders as A and the

manufacturing time matrix as B, the product can be written as:

AB = C
Therefore, C is a matrix giving the total production time on each
type of machine for each of the four orders.
The equation would no longer be true if B were written before A. In
fact, it is impossible to multiply B by A since B has three columns
while A has four rows. If the fourth row of A was disregarded,
there would then be two matrices which could be multiplied in
either order, but the results would be different:
(c)

⎡4 2 ⎤ ⎡44 216 64 ⎤ ⎡4 2 ⎤
⎢18 6 ⎥ ⎡8 38 14 ⎤ = ⎢180 876 276⎥ or ⎡8 38 14 ⎤ ⎢18 6 ⎥ = ⎡884 244 ⎤
⎢ ⎥ ⎢6 32 4 ⎥⎦ ⎢ ⎥ ⎢6 32 4 ⎥⎦ ⎢ ⎥ ⎢648 204 ⎥⎦
⎢⎣12 0⎥⎦ ⎣ ⎢⎣96 456 168 ⎥⎦ ⎣ ⎢⎣12 0⎥⎦ ⎣
UNIT 6: Matrices

The products are entirely different in the two cases. The second 65

S
product is, in fact, completely meaningless, since it includes terms Notes
such as (8 × 2) where 8 is the cutting time for sheets and 2 is an ___________________
order for pillowcases!
___________________
We showed earlier how a row vector could be multiplied by a

E
___________________
column vector to obtain a scalar product. From the rules of matrix
multiplication, we can now see that it is impossible to multiply a ___________________
row vector by another row vector or to multiply a column by ___________________
another column vector, except in the trivial case of vectors with
___________________
only a single element. Multiplying a column vector by a row vector

UP
gives a matrix instead of a scalar product: ___________________

⎡1,800⎤ ⎡7,200 3,600⎤ ___________________

⎢ 700⎥ [4 2] = ⎢ 2,800 1,400⎥
⎣ ⎦ ⎣ ⎦ ___________________

___________________
In this matrix, the elements 3,600 and 2,800 are completely
meaningless. For instance, 3600 is obtained by multiplying the
price for sheets by the order for pillowcases.

Saying that the multiplication of ordinary numbers is commutative

E-
(a×b = b×a) while the multiplication of matrices is in general non-
commutative (A×B ≠ B×A) summarizes the above result. This
means that changing the order of multiplication of two matrices
will generally change the answer. It will be seen later that there
are a few cases where changing the order does not change the
answer. A little thought will show that this can only be true for
square matrices, that is, matrices with equal numbers of row and
CC

columns; but matrix multiplication is not in general commutative

even for square matrices.

Since multiplying a vector by k means multiplying every element

by k, irrespective of whether it is a row vector or a column vector,
it is to be expected that the same rule would apply to matrices.

This is the case:

LM 4 2OP LM 4k 2kOP
kM
18 6P M18k 6k P
MM12 0PP = MM 12k 0PP
N 0 6Q N 0 6 k Q
(c)

If k = 0, one obtains a matrix in which all the elements are zero,

termed a zero matrix. A zero matrix is also obtained if two
matrices are multiplied together, one of which is a zero matrix. But
it is also possible to obtain a zero matrix as the product of two
matrices neither of which is a zero matrix:
 Quantitative Techniques for Management Applications

66
⎡ −2 1 ⎤ ⎡1 −4 ⎤ ⎡0 0 ⎤

S
Notes
Activity ⎢ 6 −3⎥ ⎢2 −8 ⎥ ⎢0 0⎥
What is the use of matrices in ⎣ ⎦ ⎣ ⎦⎣ ⎦
___________________
production planning?
___________________ Matrix addition has not been defined so far. It is possible to add
together two matrices only when both have the same number of
___________________

E
rows and the same number of columns. The sum is then obtained
___________________ simply by adding together the corresponding elements:
___________________ ⎡3 8 5 ⎤ ⎡5 30 9⎤ ⎡3 + 5 8 + 30 5 + 9 ⎤
⎢2 7 2⎥ ⎢4 25 2⎥ ⎢2 + 4 7 + 25 2 + 2⎥
___________________ ⎣ ⎦ ⎣ ⎦ ⎣ ⎦

UP
___________________ Here, the first matrix could represent the machine loading and
___________________ unloading times and the second matrix the machine running time
for the sheets and pillowcases in above example, so that the sum
___________________
would be the manufacturing time matrix already employed.
___________________
It will be seen that the rule for vector addition conforms to this
rule for matrix addition. Unlike matrix multiplication, matrix
addition is commutative; changing the order of the matrices, which
are added together, does not change the result.
E-
A few words must be added on the equality, subtraction and
division of matrices. Two matrices are said to be equal only if they
are identical; they must have the same number of rows and the
same number of columns and every element in the second matrix
must be equal to the corresponding element in the first matrix.
A matrix can be subtracted from another matrix only when both
CC

have the same number of rows and the same number of columns.
Subtraction is then simply the reverse of addition:
⎡8 38 14 ⎤ ⎡5 30 9⎤ ⎡3 8 5 ⎤
⎢6 32 4 ⎥ ⎢4 25 2⎥ ⎢ 2 7 2⎥
⎣ ⎦⎣ ⎦ ⎣ ⎦

Matrix subtraction is non-commutative, but this is to be expected

since the subtraction of ordinary numbers is also non-
commutative.
Matrix division is quickly dealt with, as it is impossible to divide a
matrix by another matrix directly. There is a round about method
which we will learn later in the unit.
(c)

Check Your Progress

Fill in the blanks:
1. Matrix subtraction is ___________________.
2. Unlike matrix multiplication, matrix addition is _______
UNIT 6: Matrices

Use of Matrices for Production Planning 67

S
Notes
Before the advent of the computer, use of matrix methods in
___________________
production planning were of theoretical interest rather than real
practical value. The position is now completely reversed. Any firm, ___________________
which has access to a computer and does not use matrix methods,

E
___________________
must be regarded as backward and inefficient. By these standards
___________________
most of the Indian firms would be!
___________________
If a firm makes m products using n different types of machines,
matrix A can represent the machine time requirements with m ___________________

UP
rows and n columns. One such matrix was shown earlier. ___________________

The total costs per minute for running each type of machine, ___________________
including both capital and labour costs, form the n elements in a
___________________
column vector. Small letters in bold type can represent such
___________________
vectors; let c be the machine-cost vector. Then Ac will be a column
vector with m elements, giving the total machine cost per unit for
each product.
If the manufacturer of sheets and pillowcases in the example above
E-
finds that the machine costs are Re 0.2 per minute for cutting, Re
0.1 for sewing and Re 0.3 for folding, then the total machine costs
per product are given by:

⎡0.2⎤
⎡8 38 14 ⎤ ⎢ ⎥ ⎡9.6⎤
⎢6 32 4 ⎥ ⎢0.1⎥ = ⎢5.6 ⎥
⎣ ⎦⎢ ⎥ ⎣ ⎦
⎣0.3⎦
CC

For most manufacturers there would, of course, be much larger

numbers of machines and products. In this section only a very
simple example can be followed through as it makes it easier to
understand the process. Computers can handle matrices with
dozens of rows and columns, each element having three or four
digits and they still use the same process.
Matrix B with m rows and q columns may represent the material
contents of the different products. With four ingredients such as
sheeting fabric, thread, labels and packing material, there would
be a four-column ingredients matrix for sheets and pillowcases,
(c)

such as:

⎡46 7 12 28⎤
⎢16 3 12 13 ⎥
⎣ ⎦
 Quantitative Techniques for Management Applications

68
The units of measurement may be different for each column, being

S
Notes
chosen to suit the nature of the ingredient. The same unit will be
___________________
used in each case when preparing the ingredient-cost vector. This
___________________ will be a column vector with q elements and may be represented by
___________________ d. Then Bd will be a column vector with m elements, giving the

E
total ingredient cost per unit for each product:
___________________
⎡2.5⎤
___________________ ⎢0.3⎥
⎡46 7 12 28⎤ ⎢ ⎥ = ⎡123.9⎤
⎢16 3 12 13 ⎥ ⎢0.1⎥ ⎢⎣ 44.7⎥⎦
___________________ ⎣ ⎦
⎢ ⎥

UP
___________________ ⎣0.2⎦

___________________ Any labour or other costs not already included in the machine-cost
and ingredient-cost vectors will be computed for each product to
___________________
form an additional cost vector e with m elements. The total cost per
___________________ unit for each product is then obtained by adding together the three
vectors each with m elements:
Ac + Bd + e = total cost vector

⎡9.6 123.9 2.0 135.5⎤

⎢5.6 44.7 1.5 51.8 ⎥
E-
⎣ ⎦
All the information about machine time requirements, material
contents and costs is collected by work study and costing staff and
the matrices A and B and vectors c, d and e are stored in the
computer. When new techniques or changed prices or wage rates
make it necessary, the matrices and vectors are brought up to date.
CC

The products Ac and Bd and the total cost vector Ac + Bd + e are

also computed and kept up to date. Depending on the pricing policy
of the firm, there may also be a selling price vector p, which is
computed from the total cost vector by adding a suitable
percentage for fixed costs and profit.
When an enquiry is received, the prices and delivery dates are
quoted by reference to the computer. If this results in an order, it is
recorded as a row vector with m elements. A row vector is
distinguished from a column vector by a distinctive mark:

x′= [4 2]
(c)

The order vector x′ may be multiplied by A to obtain a row vector

giving the production times for the order, as calculated earlier. At
the same time, x′ may be multiplied by Ac to give the total machine
cost for the order:
UNIT 6: Matrices

69
⎡9.6⎤

S
[4 2] ⎢5.6⎥ = 49.6 Notes
Activity
⎣ ⎦
Give___________________
and example of row
One would expect to obtain the same result if the production-time operation.
___________________
vector x′ A is multiplied by the machine-cost vector c and this is in
fact the case:

E
___________________

⎡0.2⎤ ___________________
[4 4 216 64 ] ⎢⎢0.1⎥⎥ = 49.6 ___________________
⎢⎣0.3⎥⎦
___________________

UP
The fact that x′A multiplied by c always gives the same result as x′ ___________________
multiplied by Ac is called the associative property of matrix
___________________
multiplication. The product can be written simply as x′Ac.
___________________
The computer stores a vector of running totals of machine-time
___________________
commitments, to which the vector x′A is added. As each order is
completed, its production times are deducted from the running
totals. The commitment for each type of machine may then be
divided by the number of machines of that type, obtaining the
E-
number of minutes and hence the number of weeks it will take to
produce all outstanding orders. This information forms the basis
for quoting delivery dates for new order and perhaps also for
planning overtime work or the purchase of additional machines.

The vector x′B is computed and similarly incorporated in the

running totals of ingredient requirements. The scalar quantity
CC

x′Bd gives the total ingredient cost of the order and may also be
added to a continuous running total if it is necessary to keep a
check on the amount of capital needed to finance work in progress.
When the order is delivered, the amount of money due is given by
x′p. This serves as a check on the invoice total.
This presentation is far from complete, but it is enough to show
how computers using matrices can keep a check on production
commitments and stocks of materials. The most important
extension necessary in most firms will take into account production
dates. The final delivery date of each order will determine the
dates by which various stages of manufacture must be completed.
(c)

Running totals of commitments will be kept by dates so that no

type of machine can be overcommitted at any stage.
Ideally, the computer will be used to print out the production
orders for each department at the appropriate times. Because t
makes all calculations extremely rapidly and keeps a complete
 Quantitative Techniques for Management Applications

70 check on all machine and material requirements, the computer will

S
Notes print each production order only when it is time to commence
___________________ production. This makes it possible to accept any rush order, which
does not conflict with existing commitments; such orders are
___________________
usually very profitable if they can be successfully handled.
___________________

E
___________________
The computer will similarly record all ingredient requirements by
dates, so that supplies are not obtained unnecessarily early. It will
___________________ be programmed to give a warning if supplies do not arrive when
___________________ they are due.

UP
___________________ Properly used, the computer is the manager’s most efficient
assistant, reminding him of any supplies that need to be chased
___________________
and warning him of any future production bottlenecks. It is a
___________________ complete waste of the computer’s powers to employ it in churning
___________________ out masses of detail which the manager has to pore through in
order to find where things may go wrong. This laborious and
inefficient management by detail must give way to management by
exception.

Let us now turn our attention to other applications of matrices.

E-
Solving Linear Equations
There are three methods that can be used to solve linear equations:
(1) Row operations (2) Matrices and (3) Determinants.
First two of these will be discuss here and third will be discuss in
next unit.
CC

Row Operations

A few hints have already been given that matrices can be used in
the solution of sets of simultaneous linear equations. But before
considering the role of matrices, it is useful to consider a technique
known as row operations.
Example 6.2
Solving linear equations by row operations is in principle the same
as solving them by elimination. The difference is that row
operations, in turn, aim systematically at a coefficient of 1 for each
unknown. For instance, in given example
(c)

3x + 11y = 180 (1)

6x + 17y = 300 (2)
The first stage in row operations is to divide the first equation by 3
in order to make the coefficient of x into 1. This equation is then
UNIT 6: Matrices

multiplied by 6 and the result subtracted from the second equation 71

S
to eliminate x from that equation. So the equations become: Notes

2 ___________________
x+3 y = 60 (3)
3 ___________________
0x + 5y = – 60 (4)

E
___________________

The next stage is to divide the second equation by – 5, in order to ___________________

make the coefficient of y equal to 1. This equation is then ___________________
multiplied by 32/3 and the result subtracted from the first equation
___________________
in order to eliminate y from that equation:

UP
___________________
x + 0y = 16 (5)
___________________
0x + y = 12 (6)
___________________
This gives the solution, x = 16, y = 12. If one is able to multiply and
___________________
subtract mentally, the whole procedure is very rapid. The other
special feature of row operations is that it is unnecessary to keep
writing the letters and the addition signs. All that is needed is to
write down the figures, keeping the zeros as in the above equations
E-
and to insert a vertical line to separate the two sides of each
equation:

⎡3 11 180 ⎤ ⎡1 32 / 3 60 ⎤ ⎡1 0 16⎤
⎢ ⎥ ⎢ ⎥ ⎢ ⎥
⎣6 17 300⎦ ⎣0 –5 −60⎦ ⎣0 1 12⎦

A set of three equations in three unknowns requires three stages.

It is necessary at each stage to write first the row, which has to be
CC

divided through to give the coefficient 1. This is the first row in the
first stage, the second row in the second stage and so on. This row
is then used to eliminate the corresponding coefficients in all the
other rows, both above and below. Hence, after the first stage the
first column of figures reads 1 0 0; after the second stage the
second column of figures reads 0 1 0, while the first columns
remains as 1 0 0 and so on.

The solution of example 3.4 by row operations reads:

⎡3 4 3 107⎤
⎢ ⎥
⎢2 5 7 148⎥
(c)

⎢⎣5 9 2 151⎥⎦

⎡1 11/ 3 1 352 / 3 ⎤
⎢ 1/ 3 ⎥
⎢0 2 5 762 / 3 ⎥
⎢0 21/ 3 −3 −271/ 3 ⎥
⎣ ⎦
 Quantitative Techniques for Management Applications

72
⎡1 0 −16 / 7 −81/ 7 ⎤

S
Notes
⎢ 1/ 7 ⎥
⎢0 1 2 326 / 7 ⎥
___________________
⎢0 0 −8 −104 ⎥
⎣ ⎦
___________________

___________________ ⎡1 0 0 16⎤

E
⎢ ⎥
___________________ ⎢0 1 0 5 ⎥
⎢⎣0 0 1 13⎥⎦
___________________
The obvious disadvantage of the method is that it introduces
___________________
fractions even when the final solution does not include

UP
___________________ any fractions. Its advantage is that it follows a strict
___________________ routine, which is always necessary if computers are to be used.

___________________ Since each row represents an equation, it is permissible to

___________________
rearrange the rows in order to get out of a difficulty. For instance,
the equations:

2x - 4y + 3z = 14 (1)
3x - 6y + 2z = 11 (2)
E-
6x - 3y + z = 2 (3)

After the first stage of row operations give:

⎡1 −2 11/ 2 7 ⎤
⎢ 1/ 2 ⎥
⎢0 0 −2 −10⎥
⎢0 −1 −½ −5 ⎥
⎣ ⎦
CC

To obtain 1 in the second position of the second row, the simplest

procedure is first to interchange the second and third rows. It
would alternatively be permissible to add the third row to the
second row. The reader should follow through both methods,
obtaining the solution x = 7, y = 3, z = 4.

Row operations can deal with a set of equations that are not all
independent.

In example 6.2 after collecting all the terms involving unknowns

on the left, the initial set of rows is:
1 1 1 1 0
−
2 4 3 4
(c)

1 3 1 1 0
−
6 4 6 8
1 1 2 1
− 0
6 4 3 4
1 1 1 5
− 0
6 4 6 8
UNIT 6: Matrices

After three stages of row operations they become: 73

S
Notes
1 0 0 −1.575 0
0 1 0 −0.75 0 ___________________

0 0 1 −1.05 0 ___________________
1 0 0 0 0

E
___________________
It is now clear that the fourth row contributes no information ___________________
other than that contained in the first three rows; in other words,
___________________
the equations were not all independent. There has been no
attempt to discard an equation arbitrarily, as done when solving ___________________

UP
this example here. If the solution of a set of equations using ___________________
determinants gives zero divided by zero for each of the
___________________
unknowns, row operations will always lead to discarding the
___________________
correct row since it will eventually produce a row of zeros. If the
equations had been contradictory, row operations would have ___________________
produced a row of zeros to the left of the vertical line with a non-
zero value on the right.
Since example 6.2 leads to three independent equations in four
E-
unknowns, it is only possible to express each unknown in terms of
one of the others. The first row in the final set of rows can be
interpreted as b = 1.575w and the other rows give the
corresponding solutions for f and h.

Using Matrices to Solve Linear Equations

Before we understand this method it is necessary for us to
CC

understand the concept of 'inverse' of a matrix.

Inverse of a Square Matrix

Example 6.3

The set of simultaneous equations used in examples 3.2 and 3.4

can be written as:

⎡3 11⎤ ⎡x ⎤ ⎡180 ⎤
⎢6 17⎥ ⎢ y ⎥ ⎢300⎥
⎣ ⎦⎣ ⎦⎣ ⎦

⎡3 4 3 ⎤ ⎡ x ⎤ ⎡107⎤
(c)

⎢2 5 7 ⎥ ⎢ y ⎥ ⎢148⎥
⎢ ⎥⎢ ⎥ ⎢ ⎥
⎣⎢5 9 2⎦⎥ ⎢⎣ z ⎦⎥ ⎣⎢151⎦⎥
It is only necessary to follow out the rules for matrix multiplication
to see that these matrix products are identical to the sets of
equations. If the symbol A is used to represent the matrix of
 Quantitative Techniques for Management Applications

74 coefficients, x to represent the vector of unknowns and B to

S
Notes represent the vector on the right-hand side, any set of
___________________ simultaneous linear equations can be expressed in the form:
___________________ Ax = B
___________________ Both A and B are known, but x is unknown. If there was a simple

E
___________________ rule for dividing B by A, the result x would be obtained. However,
it is impossible to divide a matrix directly by another matrix.
___________________
The operations of division are best understood as the
___________________
reverse of multiplication. ‘Divide B by A’ is a way of saying, 'find

UP
___________________
the matrix or vector which when pre multiplied by A will give B’.
___________________ Since matrix multiplication is in general non-commutative, it is
___________________ necessary to say ‘when pre-multiplied by A’ to indicate that A
precedes x in the above equation.
___________________
If the required matrix or vector exists, it can be found. But the
procedure is not very straightforward. It involves the new concepts
termed a unit matrix and the inverse of a matrix.
The unit matrix of order n, written In, is the square matrix with n
E-
rows and columns which has the figure 1 for each element in the
principal diagonal, the diagonal from the top left-hand corner to
the bottom right-hand corner and 0 for all other elements. Thus:

⎡1 0 0⎤
⎡1 0⎤
I2 = ⎢ I2 = ⎢⎢0 1 0⎥⎥
⎣0 1⎥⎦
⎢⎣0 0 1⎥⎦
CC

In a set of n simultaneous linear equations in n unknowns, A will

always be a square matrix. Let I be the unit matrix with the same
number of rows and columns as A, which is also the number of
elements in x, it is easy to see that:
AI=A
I A =A
I x=x
These equations can be confirmed by writing any elements one
chooses in A and x and then following the rules of matrix
(c)

multiplication. Since the product of I with any matrix or vector is

identical to the matrix or vector itself, I is sometimes called the
identity matrix.
For a given matrix A, it is usually possible to find a matrix which
when multiplied by A gives the answer I. This matrix is a sort of
UNIT 6: Matrices

reciprocal of A. The matrix which has this property is represented by 75

S
A –1 and is called the reciprocal or more commonly the inverse of A. Notes

The inverse is accordingly defined by the equation: ___________________

A-1 A = I ___________________

It can be shown that the multiplication of a square matrix and its

E
___________________
inverse always commutes, that is:
___________________
AA =I
-1
___________________
In this discussion, it has been assumed that a matrix can have only
___________________
one inverse. Although academic proofs are not the main purpose of

UP
___________________
this book, it is of interest to see how this can be proved. Let B be
any matrix, which satisfies the equation: ___________________

BA=I ___________________
Then: ___________________

B = BI = BAA-1 = IA-1 = A-1

Each step in this proof uses one of the results previously obtained,
including the associative property of matrix multiplication
E-
mentioned earlier. Since two matrices are only equal if they are
identical in all respects, this proves that B is identical to A-1 and
therefore, that there is only one inverse of A.

This proof gives a glimpse at the foothills of what mathematicians’

term matrix algebra. The usefulness of matrices lies in the fact
that one can employ them in all kinds of proofs and manipulations,
CC

simply representing each matrix by a bold letter without specifying

what its elements are or even how many rows and columns it has,
provided that one always conforms to the rules of matrix
multiplication and addition.

If the inverse of A can be found, it is easy to solve the equation:

Ax = B

It follows that :

A-1 Ax = A-1 B

Ix = A-1 B which implies x = A-1 B

(c)

So the method of solving the equations is to find the inverse of A

and then multiply this by B. This is not the same thing as dividing
B by A, but it is the nearest one can get to division of matrices.

The simplest method of finding the inverse of a square matrix is to

use row operations. Using the matrix interpretation of a set of
 Quantitative Techniques for Management Applications

76
equations, row operations involved converting the matrix A on the

S
Notes left of the vertical line into the matrix I. The effect of these
___________________ operations was to convert the vector b on the right of the vertical
line into the solution vector, which is now seen to be equal to A–1 B.
___________________
It is reasonable to deduce that the same operations will convert the
___________________ matrix I into A–1I, which is the same thing as A-1.

E
___________________
For example 3.2, the matrix of coefficients is inverted as follows:
___________________
3 11 1 0
___________________ 6 17 0 1

UP
___________________
1
1 32 / 3 0
___________________ 3
0 −5
___________________ −2 1

___________________ 17 11
−
1 0 15 15
0 1 2 1
−
5 5

As a check, the final matrix obtained on the right of the vertical

E-
line may be multiplied by the original matrix on the left to show
that it produces the unit matrix. It is then multiplied by the vector
B to obtain the solution vector:
⎡ 17 11 ⎤
⎢ − 15 15 ⎥ ⎡180 ⎤ ⎡16⎤
⎢ ⎥ ⎢300⎥ ⎢12⎥
⎢ 2 1
− ⎥ ⎣ ⎦ ⎣ ⎦
⎢⎣ 5 5 ⎥⎦
CC

In using the inverse, it is often convenient to keep fractions outside

the brackets:
1 ⎡ −17 11⎤
15 ⎢⎣ 6 − 3⎥⎦

To multiply this by the vector with elements 180 and 300 it is then
obviously easiest to start by dividing the latter numbers by 15.
For example 3.4 it is necessary to invert the matrix:
⎡3 4 3 ⎤
⎢2 5 7 ⎥
⎢ ⎥
(c)

⎢⎣5 9 2⎥⎦
UNIT 6: Matrices

Using row operations the inverse proves to be: 77

S
Notes
⎡ 53 −19 −13⎤
1 ⎢
−31 9 15 ⎥⎥ ___________________
56 ⎢
⎢⎣ 7 7 −7 ⎥⎦ ___________________

Rather large numbers become involved when the vector multiplies

E
___________________
this matrix formed by the right-hand sides of the original ___________________
equations, but eventually the number 56 cancels out:
___________________
⎡ 53 −19 −13⎤ ⎡107⎤ ⎡16⎤
1 ⎢
15 ⎥⎥ ⎢148⎥ = ⎢ 5 ⎥ ___________________
−31 9

UP
56 ⎢ ⎢ ⎥ ⎢ ⎥
⎣⎢ 7 7 −7 ⎦⎥ ⎣⎢151⎦⎥ ⎢⎣13⎦⎥ ___________________

___________________
It is unnecessary to check that the inverse was correct provided
that the solution vector is now checked in the original equations: ___________________

⎡3 4 3⎤ ⎡16⎤ ⎡107⎤ ___________________

⎢2 5 7 ⎥ ⎢ 5 ⎥ = ⎢148⎥
⎢ ⎥⎢ ⎥ ⎢ ⎥
⎢⎣5 9 2⎥⎦ ⎢⎣13⎥⎦ ⎢⎣151⎥⎦

Following this example though involves a lot of arithmetic, it must

E-
be remembered that all modern mathematical methods assume
that computers are available, so these large amounts of routine
calculations are no drawback.
If row operations are used to try to find the inverse of the matrix
from example 6.2, it will be found that the third stage gives a
fourth row consisting entirely of zeros on the left of the vertical
line. This matrix has no inverse and is known as a singular matrix.
CC

Whenever a set of equations has no unique solution, it will be

found that the coefficients form a singular matrix.

Check Your Progress

Fill in the blanks
1. Properly used, the ______________is the manager’s most
efficient assistant, reminding him of any supplies that
need to be chased and warning him of any future
production bottlenecks.
2. It is a complete ___________ of the computer’s powers to
(c)

employ it in churning out masses of detail which the

manager has to pore through in order to find where
things may go wrong.
 Quantitative Techniques for Management Applications

78
Summary

S
Notes
Matrices form one of the most powerful tools management and of
___________________
modern mathematics. They have innumerable applications in the
___________________ analysis of material and machine requirements and the solution of
___________________ problems in planning and organization. An understanding of

E
___________________
matrices is also essential for most branches of advanced
mathematics and statistics. As vectors lie at the base of matrices,
___________________
let us start by understanding them first.
___________________
Lesson End Activity

UP
___________________

___________________ A manufacturer produces three products A, B and C, which are

sold in Delhi and Calcutta. The annual sales of these products are
___________________
given below:
___________________
Product
A B C
Delhi 5000 7500 15000
Calcutta 9000 12000 8700
E-
If the sale price of the products A, B and C per unit be ` 2, 3 and 4
respectively, calculate total revenue from each centre by using
matrices.

Keywords
Matrix: An array of numbers arranged in certain numbers of rows
CC

and columns.

Rectangular Matrix: A matrix consisting of m rows and n

columns, where.

Square Matrix: If the number of rows of a matrix is equal to its

number of columns, the matrix is said to be a square matrix.

Row Matrix: A matrix having only one row.

Column Matrix: A matrix having only one column.

Singular Matrix: A matrix whose determinant is zero.

(c)

Questions for Discussion

1. Discuss with examples, different types of matrices.
2. Illustrate the use of vectors.
UNIT 6: Matrices

79
3. Given the following column vectors

S
Notes
⎡1 ⎤ ⎡ 2⎤ ⎡ 2⎤
A = ⎢⎢2⎥⎥ , B = ⎢-1⎥ , C = ⎢-2 ⎥ ___________________
⎢ ⎥ ⎢ ⎥
⎣⎢3 ⎦⎥ ⎣⎢ 3⎦⎥ ⎣⎢ 1⎦⎥ ___________________

E
___________________
Find (i) A+B, (ii) A+B+C, (iii) 2A-B+C
___________________
⎡2 3⎤ ⎡1 2 3 ⎤
4. If A = A = ⎢ ⎥ and B = ⎢ ⎥
⎣3 7⎦ ⎣3 2 4 ⎦ ___________________

___________________
Then find AB.

UP
___________________
5. For square matrices A and B, expand (A+B) (A-B) and (A-B)
(A+B). When will these products be equal? ___________________

___________________
6. Nahar Chemical Mills produces three varieties of base oil,
Super fine Grade (A grade), fine grade (B grade) and coarse ___________________
grade (C grade). The total annual sales in lacs of rupees of
these products for the year 1999 and 2000 in the four cities is
given below, find the total sales of three varieties of base oil
for two years.
E-
For the year 1999

City Product Calcutta Mumbai Chennai Delhi

Superfine base oil (A grade) 30 16 12 24
Fine base oil (B grade) 10 48 14 16
Coarse base oil (C grade) 16 8 62 12

For the year 2000

CC

City Product Calcutta Mumbai Chennai Delhi

Superfine base oil (A grade) 34 20 10 14
Fine base oil (B grade) 10 44 22 8
Coarse base oil (C grade) 26 12 78 10

7. A 2T oil manufacturer produces three products A, B, C which

he sells in the market. Annual sale volumes are indicated as
follows:
Market Products
A B C
I 8,000 10,000 15,000
II 10,000 2,000 20,000
(c)

If the unit sale price of A, B and C are ` 2.25, 1.50 and ` 1.25
respectively, find the total revenue in each market with the
help of matrices. (ii) If the unit costs of above three products
are ` 1.60, ` 1.20 and ` 0.90 respectively, find the gross profit
with the help of matrices.
 Quantitative Techniques for Management Applications

80
8. In a development plan of a cracker complex, a contractor has

S
Notes
taken a contract to construct certain buildings for which he
___________________
needs building materials like stone, sand, etc. There are three
___________________ firms A, B, C that can supply him these materials. At one time
___________________ these firms A, B, C supplied him 40, 35 and 25 truck loads of

E
stones and 10, 5 and 8 truck loads of sand respectively. If the
___________________
cost of one truck load stone and sand are ` 1,200 and ` 500
___________________ respectively then find the total amount paid by the contractor
___________________
to each of these firms A, B and C separately.

UP
___________________ 9. HEG Ltd. maintains the records of the daily cost C of
operating the “stress relieving furnaces division",
___________________
which is a linear function of the number of incoming
___________________ electrodes 1 and outgoing electrodes P, plus a fixed cost a, i.e.
___________________ C = a + bp = d1
Given the following data for 3 days find the values of a, b, and
d by setting up a linear system of equations and using the
matrix inverse.
No. of Incoming No. of Outgoing
E-
Cost Electrodes to stress electrodes from
Day
(in `) relieving furnaces stress Relieving
division furnaces division
1 6,950 40 10
2. 6,725 35 9
3. 7,100 40 12

10. Mr. Bhattacharya has retired from service in 1999 from IOC.
CC

He received ` 14 lacs as the provident fund and retirement

benefits from the company. He decides to invest a sum of `
40,000 in three different stocks that yield 10%, 12 % and 15%
respectively. The income from the third stock (which yields
15%) is twice the income from the first stock (which yields
10%). After one year, Mr. Bhattacharya earned an income of `
5000 from his investments. What is the amount that he has
invested in each type of stock.

11. Robin Singh & Company Ltd. stocks lubes of Castrol brand
and Mak brand. The matrix of transition probabilities of the
lubes is shown below:
(c)

Castrol Mak
Castrol 0.9 0.1
Mak 0.3 0.7

Determine the market share of each of the brand in

equilibrium position.
UNIT 6: Matrices

Further Readings 81

S
Notes
Books ___________________

R S Bhardwaj, Mathematics for Economics and Business, Excel ___________________

Books, New Delhi, 2005

E
___________________

D C Sanchethi and V K Kapoor, Business Mathematics ___________________

Sivayya and Sathya Rao, An Introduction to Business Mathematics ___________________

___________________
Web Readings

UP
___________________
www.textbooksonline.tn.nic.in
___________________
www.mathbusiness.com
___________________

___________________
E-
CC
(c)
 Quantitative Techniques for Management Applications

82

S
Notes

___________________

___________________

___________________

E
___________________

___________________

___________________

UP
___________________

___________________

___________________

___________________
E-
CC
(c)
UNIT 7: Determinants

Unit 7
83

S
Activity
Notes
Differentiate matrices and

Determinants
___________________
determinants
___________________

E
___________________
Objectives
___________________
After completion of this unit, the students will be aware of the following
topics: ___________________

\ Determinants ___________________

UP
\ Solution of Linear Equations using Determinants. ___________________
\ Properties of Determinants
___________________

___________________
Introduction
___________________
In this unit, we will discuss the concepts of determinants and their
importance in solving real world problems of business. While a
matrix is an array of numbers arranged into certain number of
rows and columns, a determinant is a scalar associated with a
E-
square matrix. Unlike scalars, the basic operations such as
addition, subtraction and multiplication can be performed only if
certain conditions are satisfied by the participating matrices. Like
scalars division of one matrix by another is not defined.

Solving linear Equation by the use of Determinants

CC

Any pair of simultaneous equations in two unknowns can be

written in the form:
a 1x + b 1y = h 1 (1)

a 2x + b 2y = h 2 (2)

The symbols, other than x and y, represent known quantities and

the solution by elimination can be carried out in the same way as
done earlier. The first step is to multiply the first equation
throughout by a2 and the second equation throughout by a1.

a 1 a 2x + a 2 b 1y = a 2 h 1 (3)
(c)

a 1 a 2x + a 1 b 2 y = a 1 h 2 (4)

Then, subtract the first equation from the second in order to

eliminate x:
a 1b 2y - a 2 b 1y = a 1 h 2 - a 2 h 1
 Quantitative Techniques for Management Applications

84
a1 h2 − a2 h1
(5)

S
Notes y=
a1 b2 − a2 b1
___________________
Substituting the value of y in the first equation and rearranging
___________________
we find that
___________________

E
h1 b2 − h2 b1
___________________ x= (6)
a1 b2 − a2 b1
___________________
These two formulae are called the general formulae for the solution
___________________
of any pair of simultaneous linear equations.

UP
___________________
The next important step is to introduce a particular way of
___________________
representing this solution so that it is easy to remember. The
___________________ method adopted is to write:
___________________ a1 b1
= a1 b2 − a2 b1
a2 b2

The left-hand side of this equation is known as a determinant and

the symbols between the two vertical lines are termed the
E-
elements of the determinant. Any set of four symbols or numbers
arranged in this way between vertical lines is always interpreted
as the difference of two products in the order shown. It should also
be noted that this is a special kind of equation called an identity
and is true for any values or symbols. Therefore, it is a general
mathematical truth or definition rather than a statement about
particular values or symbols. An identity is often represented by a
CC

triple hyphen in place of the double hyphen for an ordinary

equation.
Two simple properties of determinants are immediately obvious.
The value of the determinant is unchanged if the first row becomes
the first column and the second row becomes the second column;
but the value is multiplied by – 1 if the first and second rows are
interchanged.
These two properties can be represented by the identities:

a1 b1 a1 a2
≡
a2 b2 b1 b2
(c)

a1 b1 a b2
≡ −1 2
a2 b2 a1 b1

These identities can be proved by multiplying both sides in

accordance with the original definition of a determinant. This is
UNIT 7: Determinants

known as expanding the determinants. There is a third simple 85

S
property which can be similarly proved; if any multiple of one row Notes
or column is added to the other row or to the other column ___________________
respectively, the value of the determinant is unchanged:
___________________
a1 b1 (a1 + pa 2 ) (b2 + pb2 ) (a1 + qb1 )b1
≡ ≡

E
___________________
a2 b2 a2 b2 (a2 + qb2 )b2
___________________
whatever the values of p or q. ___________________

So determinants have properties which can be proved and utilized ___________________

UP
in solving linear equations. So far the determinants considered
___________________
have two rows and two columns and called second order
determinants, used in solving a pair of simultaneous linear ___________________
equations in two unknowns. Once the method of solution is ___________________
understood, it can be extended to larger systems of equations
___________________
where the practical merits of the method are much more apparent.

Given the two equations at the beginning of this section, the first
step is to take the four coefficients on the left-hand sides and write
them as a determinant. This forms the denominator of the formula
E-
for x. The numerator is the same determinant with the coefficients
of x replaced by the column of values from the right-hand sides of
the equations. So the formula is:

h1 b1
h2 b2
x=
a1 b1
CC

a2 b2

By expanding these determinants, it can be seen that this is the

same solution as that obtained by elimination. Since the setting-up
of the determinants is a routine procedure and their evaluation is
also routine, the original objective has been achieved. The
equations can now be solved using a computer, which is able to
carry out routine calculations but unable to think for itself.

We still need to point out that the value of y is represented in the

same way as the value of x. The denominator is the same and the
numerator is obtained by replacing the coefficients of y by the
column of values on the right-hand sides of the equations:
(c)

a1 h1
a2 h2
y=
a1 b1
a2 b2
 Quantitative Techniques for Management Applications

86
Check Your Progress

S
Notes
Activity
What is the best method to Fill in the blanks
___________________
solve linear equations?
___________________
1. The left-hand side of any equation is known as a
__________.
___________________

E
2. The symbols between the two vertical lines are termed
___________________
the ____________ of the determinant.
___________________

___________________ The Best Method of Solving Linear Equations

UP
___________________
Three different systematic techniques have been presented for the
___________________ solution of sets of simultaneous linear equations: determinants,
___________________ row operations and matrices. It is reasonable to ask which is the
best method.
___________________
The method of determinants is usually the quickest if one requires
the value of a few unknowns. This often applies when the
equations arise from problems in probability. The determinant of
the coefficients on the left-hand side of the equations should
E-
always be evaluated if there is a doubt whether a set of equations
has a unique solution.
The quickest way to obtain a complete solution for all the unknowns
is usually by row operations. If the method of determinants gives zero
divided by zero for each of the unknowns, only row operations will
provide the solution, which expresses the unknown in terms of one
CC

another, there being no unique solution. Row operations are

obviously quicker than matrix methods, since they obtain the
whole solution with no more work than is involved in inverting the
matrix.
Matrix methods are extremely valuable if there are a large number
of different sets of equations, all of which have the same matrix of
coefficients. For instance, it might be desired in example 6.2 to
consider all the possible product mixes obtainable by adding or
subtracting one or two machines of each of the four kinds. There
would then be several hundred possible combinations of machines,
each giving rise to a set of equations with the same matrix of
(c)

coefficients. The inverse of this matrix can be found and then each
set of equations is solved simply by multiplying this inverse matrix
by the appropriate vector.
UNIT 7: Determinants

87
All three methods can be applied using computer, but in practice

S
Notes
computer programs tend to prefer matrix methods because of the Activity
wider applications of matrices. What are higher order
___________________
matrices?
It will be observed that a set of equations which has no unique ___________________

solution always gives a determinant of value zero and always

E
___________________
provides a singular matrix. It is customary to speak of ‘the
___________________
determinant of a matrix’ provided that it is a square matrix. A
___________________
singular matrix may be defined either as a matrix which has no
inverse or as a matrix whose determinant is zero. ___________________

UP
___________________
Check Your Progress
___________________
Fill in the blanks:
___________________
1. It will be observed that a set of equations which has no
___________________
unique solution always gives a determinant of value
_______ and always provides a ___________matrix.
2. A set of equations which has no unique solution always
gives a __________ matrix
E-
Higher Order Determinants
The practical importance of determinants increases when it is
necessary to solve large numbers of equations with
correspondingly large number of unknowns. It would be very
convenient if the solution of a large set of linear equations could be
written out in determinant form by the same routine as has been
CC

described for a pair of equations. If we followed this procedure, the

solution of the equations:
a1x + b1y + c1z = h1 (1)

a2x + b2y + c2z = h2 (2)

a3x + b3y+ c3z = h3 (3)

ought to be:

h1 b1 c1
h 2 b2 c 2
(c)

h 3 b3 c 3
x=
a1 b1 c1
a 2 b2 c 2
a 3 b3 c 3

But is it that simple?

 Quantitative Techniques for Management Applications

88
Until now, a determinant with three rows and three columns,

S
Notes
termed a Third Order determinant, has not been defined; there
___________________
are, therefore, no rules for evaluating it. Since the purpose of
___________________ determinants is to enable the solution of a set of equations to be
___________________ determined, the rules for evaluating determinants of higher order

E
are designed to fulfil this purpose.
___________________
The value of x in the solution of the given set of equations is found
___________________
by elimination to be:
___________________
h1 (b2 c3 – b3 c2 ) – b1 (h 2 c3 – h3 c2 ) + c1 (h 2 b3 – h3 b2 )

UP
___________________
a1 (b2 c3 – b3 c2 ) – b1 (a2 c3 – a3 c2 ) + c1 (a2 b3 – a3 b2 )
___________________
The terms have here been arranged so that all the expression in
___________________ brackets can be expressed as second order determinants. It can
___________________ now be seen that this will be the same formula as given previously.
So the rule for evaluating a third order determinant is:

a1 b1 c1 b2 c2 a2 c2 a2 b2
a2 b2 c 2 ≡ a1 − b1 + c1
E-
a3 b3 c3 b3 c3 a3 c3 a3 b3

The values of y and z are obtained by expressing the information in

the original equations in the form of determinants by the same
routine as for a pair of equations; that is, the denominator is the
same as in the formula for x, and the column of values from the
right hand sides of the equations replaces the coefficients of y or z
CC

in the determinants forming the respective numerators.

a1 h1 c1 a1 b1 h1
a 2 h2 c2 a 2 b2 h 2
a3 h3 c3 a 3 b3 h 3
y= z=
a1 b1 c1 a1 b1 c1
a 2 b2 c 2 a 2 b2 c 2
a 3 b3 c 3 a 3 b3 c 3

The same principles can be applied to solve larger sets of

equations, the determinants being set up by exactly the same
routine. A determinant always has the same number of rows as
(c)

columns, corresponding to the number of equations and unknowns

that have to be solved.
The rule for evaluating a determinant of n rows and columns is to
take the n elements in the first row and multiply each by the
smaller determinant obtained by missing out the
UNIT 7: Determinants

first row and the column which contains the elements in question. 89

S
Thus, the first element is multiplied by the original determinant Notes
Activity
reduced by the first row and the first column; Discuss the main properties of
___________________
the second element is multiplied by the original determinant determinants.
___________________
reduced by the first row and the second column and so on. Finally,

E
the products are collected together by adding all ___________________

the odd ones and subtracting all the even ones. So a fourth order ___________________
determinant is expanded as:
___________________
a1 b1 c1 d1
b2 c2 d2 a2 c3 d2 ___________________
a2 b2 c2 d2

UP
= a1 b3 c3 d 3 − b1 a3 c3 d3
a3 b3 c3 d3 ___________________
b4 c4 d4 a4 c4 d4
a4 b4 c4 d4 ___________________

a2 b2 d2 a2 b2 c2 ___________________

+ c1 a 3 b3 d3 − d1 a3 b3 c3 ___________________
a4 b4 d4 a4 b4 c4

Third order determinants, which occur in this expansion are

termed as minors of the original determinant. If the minor is
E-
prefaced by the appropriate sign, it is termed as the
co-factor of the element by which it is multiplied. Thus, the
cofactor of b1 is:

a2 c2 d 2
− a3 c3 d 3
a4 c4 d 4
CC

Check Your Progress

Fill in the blanks:
1. The rule for evaluating a determinant of n rows and
columns is to take the n elements in the first row and
multiply each by the smaller determinant obtained by
__________out the first row and the column which
contains the elements in question.
2. The practical importance of determinants increases
when it is required to solve _________ no of equations.
(c)

Main Properties of Determinants

Unless a computer is being used, a larger determinant is hardly
ever multiplied out as it stands. It is much easier to start off by
simplifying it, using the principal properties of determinants to
 Quantitative Techniques for Management Applications

90 reduce large numbers to small numbers or to zero wherever

S
Notes possible.
___________________
Three of the principal properties have already been mentioned
___________________ when discussing second order determinants and it can be proved
___________________ that they remain true for determinants of higher order. The main

E
properties of the determinants can be summarized as under:
___________________
1. The value of the determinant is unchanged if the rows and
___________________
columns are interchanged with each other.
___________________ 2. The value of the determinant is multiplied by –1 if any row is

UP
___________________ interchanged with any other row. It follows from these first
___________________
two properties that the value of the determinant is multiplied
by – 1 if any two columns are interchanged.
___________________
3. If any multiple of any row or column is added (or subtracted)
___________________
to any other row or any other column respectively, the value of
the determinant is unchanged. However, one cannot add a
multiple of a row to a column, or vice versa.
4. If any row or column has a factor common to all its elements,
then this factor may be divided out. For instance, it can be
E-
seen by expanding both sides that:
a1 b1 c1 a1 b1 c1
pa2 pb2 pc 2 ≡ p a2 b2 c 2
a3 b3 c3 a3 b3 c3

5. It follows from (4) that if any determinant has a row of zeros or

CC

a column of zeros, the value of the determinant is zero because

then you can take out 0 common and multiply the whole
determinant with it resulting in a zero.
6. It follows from (3) and (5) that if any row is identical to any
other row or a multiple of any other row (or if any column is
identical to any other column or a multiple of any other
column) then the value of the determinant is zero.
We would like to illustrate the use of these six properties
before we introduce the final three properties.

Example
(c)

1 0
1 0

The value is obviously zero, using the properties (5) or (6)

discussed above.
UNIT 7: Determinants

Example 91

S
Notes
3 6
is zero, using (6)
2 4 ___________________

___________________
The first row is 1½ times the second row.

E
___________________
Example
___________________
4 7 4 7
is equal to u sin g (3) ___________________
11 13 3 −1
___________________
By subtracting twice the first row from the second row, the

UP
___________________
numbers are made smaller and so easier to manipulate.
___________________
In evaluating a large determinant, property (4) is first applied
wherever possible. Then property (3) is used where possible to ___________________
make the numbers small and in particular to obtain as many ___________________
zeros as possible.

Example
12 17 36 19 12 17 36 19
E-
−11 13 28 14 −11 13 28 14
=3
9 15 12 6 3 5 4 2
13 −5 −8 11 13 −5 −8 11
12 17 9 19 12 −1 9 1
−11 13 7 14 −11 −1 7 0
= 12 = 12
3 5 1 2 3 3 1 0
CC

13 −5 −2 11 13 −1 −2 15

13 −1 9 1 12 −1 9 1
−10 −1 7 0 −10 −1 7 0
= 12 = 12
0 3 1 0 0 3 1 0
14 −1 −2 15 −1 −1 −2 15

The numbers have now been reduced quite a lot and there is a
row which includes two zeros. The next step is to make a third
zero in this row. The third row is then interchanged with the
second row and again interchanged with the first row. The fourth
order determinant then reduces to a single third order
(c)

determinant.

12 −28 9 1 12 −28 9 1
−10 −22 7 0 0 0 1 0
12 = − 12
0 0 1 0 −10 −22 7 0
−1 5 −2 15 −1 5 −2 15
 Quantitative Techniques for Management Applications

92
0 0 1 0

S
Notes 12 −28 1
12 −28 9 1
___________________ = 12 = 12 −10 −22 0
−10 −22 7 0
___________________ −1 5 15
−1 5 −2 15
___________________

E
___________________
The same procedure can now be applied to this smaller
___________________ determinant. A second zero is obtained in the final column,
this column is moved to the first column by applying property
___________________
(2) twice and property (1) is then applied before the final

UP
___________________ evaluation.
___________________ 12 −28 1 12 −28 1 12 1 −28
___________________ 24 −5 −11 0 = 24 −5 −11 0 = −24 −5 0 −11
___________________ −1 5 15 −181 425 0 −181 0 425

1 12 −28 1 0 0
= 24 0 −5 −11 = 24 12 −5 −181
0 −181 425 −28 −11 425
E-
−5 −181
= 24
−11 425

= 24 [(- 5 × 425) – (- 11 × – 181)]

= 24 (-2125 – 1991) = 24 × – 4,116

CC

= - 98,784
This example is a long one, since the original determinant was
quite a large one. It should also be remembered that it is much
easier to apply the rules for simplifying and evaluating
determinants oneself than to follow the steps which someone
else has chosen. Often there is more than one route to the
answer and the route adopted may be a matter of choice.
The last three of the principal properties of determinants
combine together some of the earlier properties, enabling one
to amalgamate two or three steps into one.
(c)

7. If any row is moved up or down an even number of rows, the

value of the determinant remains unchanged. If it is moved an
odd number of rows, the value of the determinant is
multiplied by –1. Similarly, if any column is moved an even
number of columns, the value is unchanged; if it is moved an
odd number of columns, the value is multiplied by – 1.
UNIT 7: Determinants

93
12 −28 9 1 0 0 1 0

S
Notes
−10 −22 7 0 12 −28 9 1
= −1
0 0 1 0 −10 −22 7 0 ___________________

−1 5 −2 15 −1 5 −2 15 ___________________

E
___________________
12 −28 1 1 12 −28
−5 −11 0 = 0 −5 −11 ___________________

−181 425 0 0 −181 425 ___________________

8. A determinant can be expanded using the elements in the first ___________________

UP
column instead of the elements in the first row. Each element ___________________
is multiplied by its minor, the smaller determinant obtained ___________________
by deleting the first column and the row which contains the
___________________
element. The products are then alternately added and
subtracted in the same way as when expanding by the first ___________________

row.

12 −28 1
−11 10 −28 1 −28 1
−5 −11 0 = 12 − ( −5) + (−1)
5 15 5 15 −11 0
E-
−1 5 15

9. If there is any row or column in which all the elements, except

one, are zero, the determinant is equal to the product of that
one non-zero element and its cofactor. For an element in the jth
row and the kth column, the cofactor is obtained by deleting the
jth row and kth column to obtain the minor and then
CC

multiplying by –1 if (j + k) is odd but leaving it unchanged if (j

+ k) is even.

12 −28 9 1
−28 1
12
−10 −22 7 0
= −10 −22 0 (Positive since j = 3 and k=3)
0 0 1 0
−1 5 15
−1 5 −2 15

12 −28 1
−5 −11
−5 −11 0 = (Positive since j = 1 and k = 3)
−181 425
−181 425 0
(c)

2 7 0 5
2 7 5
−1 3 0 4
= − 7 −1 3 4 (Negative since j=4 and k=3)
4 2 0 −2
4 2 −2
5 −4 7 1
 Quantitative Techniques for Management Applications

94
All these properties of determinants can be proved, but the

S
Notes
Activity
proofs are of interest only to mathematicians. They are all
Discuss the applications of
___________________
determinants in management. intuitively reasonable and the manager who needs to use
___________________ determinants should be content to accept that they are valid
___________________ for determinants of any order.

E
___________________ Check Your Progress
___________________ Fill in the blanks:

___________________ 1. A determinant can be expanded using the elements in

UP
the first column instead of the elements in
___________________
the__________.
___________________
2. If any row is moved up or down an even no. of rows, the
___________________
value of determinant remains ______________ .
___________________

Applications in Management

Managerial Problems Involving Determinants

E-
The systematic nature of determinants makes it possible to use
them in solving sets of simultaneous linear equations with the aid
of a computer and so take the drudgery out of mathematics. It is
still necessary to understand the nature of determinants and their
principal properties in order to know what the computer is doing
and why it may sometimes fail to produce a solution. It is also an
advantage to be able to solve fairly simple sets of equations
CC

without the complications of computers, either by calculation on

papers or by using a calculator.
An obvious managerial application of determinants is in the type of
problem illustrated below, where it is required to find the correct
product mix in order to make full use of machine capacities.
Example 7.1
In a petroleum engineering workshop there are seven machines for
drilling, two for turning, three for milling and one for grinding.
Four types of brackets are made. Type A is found by work study to
require 7 minutes drilling, 3 minutes turning, 21/2 minutes milling,
(c)

and 11/2 minutes grinding, and the corresponding times in minutes

for the other types are: B: 5, 0, 1 1/2, 1/2; C: 14, 6, 9, 3 1/2; D: 26, 9, 11,
11/2. How many of each type of brackets should be produced per
hour in order to keep all the machines fully occupied?
UNIT 7: Determinants

95
Solution

S
Notes
The four equations could be set up in the same way as done earlier,
___________________
each equation representing the total minutes of work per hour on a
particular type of machine. The numbers on the right-hand sides ___________________

would therefore be 420, 120, 180, and 60 respectively. The

E
___________________
coefficients on the left-hand sides would form the determinant
___________________
constituting the denominator in the solution. This can be written
___________________
out directly from the information given in the question and then
evaluated: ___________________

UP
___________________
7 5 14 26
3 0 6 9 7 5 14 26 0 5 0 5 ___________________

1 1 31 0 2 3 3 1 0 2 3
9 11 = = ___________________
2 1
2 2 45 3 18 22 4 0 3 8 7
___________________
1 1 1 1 3 1 7 3 0 1 1 −6
1 3 1
2 2 2 2

The operations has been as follows. From second row 3 and from
3rd and 4th row ½ has been taken out as common. Then, 1 in the
E-
second determinant the second row multiplied by 7, 5 and 3
respectively is subtracted from 1st, 3rd and 4th row respectively.
In the fifth determinant first column is subtracted from third
column.

5 0 5 1 0 1
3 −15
=− 3 8 7 = 3 8 7
4 4
CC

1 1 −6 1 1 −6

1 0 0
−15
= 3 8 4
4
1 1 −7

−15 8 4 2 1
= = −15
4 1 −7 1 −7
= –15 (–14 –1) = 225

The numerator for type A appears more formidable at first, but can
be quickly simplified:
(c)

420 5 14 26 7 5 14 26 7 5 14 26
120 0 6 9 2 0 6 9 2 0 6 9
= 60 = 50
180 3 9 11 3 3 9 11 6 3 18 22
60 1 3 1 1 1 3 1 2 1 7 3
 Quantitative Techniques for Management Applications

96
−3 0 −21 11

S
Notes
Activity −3 −21 11
2 0 6 9
What is brand switching?
___________________ = 15 = 15 2 6 9
0 0 −3 13
___________________ 0 −3 13
2 1 7 3
___________________

E
−3 −7 11
___________________
= 45 2 2 9
___________________ 0 −1 13
___________________
−3 −7 16

UP
___________________ −3 −80
= 45 2 2 35 = 45
2 35
___________________ 0 −1 9
___________________
−3 −16
___________________ = 225
2 7
= 225(–21 + 32) = 225 × 11
Since the denominator is 225, the number of brackets per hour of
type A in the solution is 11. There are no new problems involved in
E-
finding the solution for the other three types of bracket and so this
task is left for you. You should note that it has not been necessary
either to introduce symbols for the four unknowns or to write out
the four equations.
Most managerial applications of determinants are as
straightforward as this example.
CC

Determinants are useful when you have to solve a number of

equations, but are difficult to use when you have to look at
computerization of the firm’s operations. We then turn to matrices,
to see how these can be used to solve managerial problems.

Brand Switching and Markov Chains

Let us take up a managerial application of matrices, which is of
theoretical as well as practical interest. This is mostly used by
advertising agencies and big companies in brand management.
Example 7.2
(c)

Three brands of detergent share the market, 40% of customers

buying brand A, 50% brand B, and 10% brand C. Each week there
are changes in the customers’ choices. Of those who bought brand
A previous week, 50% buy it again, but 15% change to brand B and
35% to brand C. Of those who bought brand B, 60% buy it again,
UNIT 7: Determinants

10% buy brand A and 30% buy brand C. Of those who bought 97

S
brand C, 85% buy it again, 5% buy brand A and 10% buy brand B. Notes
What proportion of the market will each of the three brands ___________________
eventually hold?
___________________
Solution:

E
___________________
It is simplest to express the brand switching percentage as
___________________
decimals, keeping percentage figures for the market shares. The
___________________
change in market shares in the first week can be obtained as the
product of a matrix representing the brand switching and a vector ___________________

UP
representing the initial market shares: ___________________

⎡0.50 0.10 0.05⎤ ⎡40⎤ ⎡25.5⎤ ___________________

⎢0.15 0.60 0.10⎥ ⎢50 ⎥ = ⎢37.0⎥
⎢ ⎥ ⎢ ⎥ ⎢ ⎥ ___________________
⎣⎢0.35 0.30 0.85⎦⎥ ⎢⎣10 ⎦⎥ ⎣⎢37.5⎦⎥
___________________
It will easily be seen that the terms involved in this product, (0.50
× 40), (0.10 × 50), etc., are the correct calculations from the
information given in the example in order to obtain the new
market shares. In this type of model, each of the matrices adds up
E-
to 1 and the elements in each vector total 100.
For the following week, the new market-share vector must
multiply the same brand switching matrix:

⎡0.50 0.10 0.05⎤ ⎡25.5⎤ ⎡18.325 ⎤

⎢0.15 0.60 0.10⎥ ⎢37.0⎥ = ⎢29.775⎥
⎢ ⎥ ⎢ ⎥ ⎢ ⎥
CC

⎢⎣0.35 0.30 0.85⎥⎦ ⎢⎣37.5⎥⎦ ⎢⎣51.900 ⎥⎦

As could have been guessed from the original information, brand C

is getting a steadily larger share of the market. It cannot, however,
obtain a monopoly. A little arithmetic will show that if brand C has
80% of the market one-week it cannot have more than 75% the
next week. Clearly its eventual share will be somewhere between
51.9% and 75%, but obviously a direct method of finding the
eventual share is desirable.
Let the brand switching matrix be M and the successive market-
share vector a, b, c… so that the above two equations can be
written as:
(c)

Ma = b
Mb = c
 Quantitative Techniques for Management Applications

98
A vector x will represent the final market shares such that pre-

S
Notes
multiplying it by the brand switching matrix produces an
___________________ unchanged result. That is:
___________________ Mx = x
___________________ Using I for the unit matrix and 0 for a zero matrix or vector, this

E
___________________ can be rearranged:

___________________ Mx = Ix

___________________ Mx – Ix = 0

UP
___________________ (M – I) x = 0

___________________ The fact that the last equation is equivalent to the preceding one
depends on the property of matrix multiplication termed
___________________
distributive. The full summary of the properties of matrix
___________________ multiplication is that it is associative and distributive but not, in
general, commutative. The latter property makes it essential to
place M and I before x in both equations to be sure that they are
equivalent.
E-
Since M must be a square matrix and I is chosen to have the same
number of rows and columns as M, there is no difficulty in finding
the matrix (M –I):

⎡0.50 0.10 0.05⎤ ⎡1 0 0⎤ ⎡ −0.50 0.10 0.05 ⎤

⎢0.15 0.60 0.10⎥ − ⎢0 1 0⎥ = ⎢ 0.15 −0.60 0.10 ⎥
⎢ ⎥ ⎢ ⎥ ⎢ ⎥
⎢⎣0.35 0.30 0.85⎥⎦ ⎢⎣0 0 1 ⎥⎦ ⎢⎣ 0.35 0.30 −0.15⎥⎦
CC

One cannot find x by obtaining the inverse of (M-I), because it is a

singular matrix. Since each column of M totals 1, each column of
(M-1) must total zero. The determinant of this matrix must
therefore be equal to zero, applying properties (3) and (5) of
determinants section in adding the second and subsequent rows to
the first row. It follows that (M-I) will always be a singular matrix
in this type of problem.
The equations represented by the matrix equation are clearly not
all independent. When it has been stated what proportions of
previous customers are retained or gained by all brands except the
(c)

last, the equations giving the market share of the last brand is
made up of what is left.
There is an additional fact not included in the matrix equation.
Expressing the market shares as percentages, all the elements in
UNIT 7: Determinants

the vector x must total 100. This gives an additional row to permit 99

S
a unique solution to be found by row operations: Notes

___________________
−0.50 0.10 0.05 0
0.15 −0.40 0.10 0 ___________________

0.35 0.30 −0.15 0

E
___________________
1 1 1 1000
___________________
In the process of solution, one of the rows will become a row of ___________________
zeros and the remaining rows will give a unique solution. It is left
___________________
to you to confirm that the final shares are respectively 12/109, 23/109

UP
and 74/109, or approximately 11.0%, 21.1% and 67.9%. ___________________

___________________
Forming it into a vector and pre-multiplying this vector by the
brand switching matrix can check the solution. It will be noted ___________________
that the information about the initial market shares is not used in ___________________
finding the solution. The final market shares will be exactly the
same whatever shares the brands started with.
As a practical application, the brand switching example has two
drawbacks. First, repeated matrix multiplications will eventually
E-
involve fractions of customers, which is impossible. Second, it is
highly improbable that there will be a persistent pattern of brand
switching; either customer will become less inclined to switch
brands or the pattern of switching will be disrupted by special
sales campaigns or dynamic market forces.
However, the model has many other applications more realistic
CC

than brand switching, particularly when the discussion is

transferred from proportions to probabilities. Where an operator
attends several machines which are subject to random stoppages
at differing average frequencies, so that two or more may be
stopped at the same time causing machine interference, the
technique here discussed enables the average productivity of each
machine to be accurately calculated. It can also be used to
calculate average stock levels and the probability of running out of
stock when demand and supply are random and so assist in finding
the optimum stockholding policy. The model has the impressive
title ergodic Markov Chains.
(c)

You would have noticed that one or two theoretical difficulties have
been ignored in presenting the technique. How can it be proved
that the market shares will settle down to a stable vector? Is it
always permissible to use an equation, which assumes the
existence of the stable vector?
 Quantitative Techniques for Management Applications

100
The techniques are, in fact, valid for the types of problems, which

S
Notes
have been discussed. But it would be very unwise to apply it
___________________
unthinkingly in quite different situations, where not all the
___________________ elements in the matrix are positive. Pure mathematicians go
___________________ deeply into all questions of validity and it is necessary to seek their

E
guidance whenever the appropriateness of certain techniques is in
___________________
doubt.
___________________
Several examples of problems involving sets of simultaneous linear
___________________ equations were discussed in unit 3, which could be solved either by

UP
___________________ elimination or by substitution. However, all these methods require
___________________
certain amount of ingenuity and a great deal of calculation.
Attempting to solve ten or twelve simultaneous equations by
___________________
elimination would be a task not meant for the fainthearted and
___________________ usually managers do not have the time to do it. Yet in many
important applications, such as finding the product mix that will
keep a number of different types of machines fully occupied, it is
not unusual to have large numbers of equations and unknowns. If
this amount of detail cannot be avoided, it is reasonable to look for
E-
a routine method of solving equations which is so automatic that a
computer can be employed to find the solution. Such routine
methods, involving determinants and matrices, are discussed here.
Keeping all the machines fully occupied is not necessarily the
manager’s main objective. There may be other product mixes which
leave some machines partly unoccupied but yield a greater total
CC

profit. The manager’s usual objective is to find the product mix

which will maximize the total profit without exceeding the
capacities of the machines.

Check Your Progress

Fill in the blanks
1. The manager’s usual objective is to find the product mix
which will __________ the total profit without exceeding
the capacities of the machines.
2. The equations represented by the matrix equation are
(c)

clearly ________all independent.

Summary
The method of determinants is usually the quickest if one requires
the value of a few unknowns. This often applies when the
UNIT 7: Determinants

equations arise from problems in probability. The determinant of 101

S
the coefficients on the left-hand side of the equations should Notes
always be evaluated if there is a doubt whether a set of equations ___________________
has a unique solution
___________________

E
___________________
Lesson End Activity
___________________
A manufacturer produces three products A, B and C, which he sells
___________________
in two markets. Annual sales volumes are indicated as follows:
Market Product A Product B Product C ___________________

UP
I 10000 2000 18000 ___________________
II 6000 20000 8000
___________________
If each unit of A, B and C is sold at ` 2.50, 1.25 and 1.50 ___________________
respectively, find revenue from each market.
___________________

Keywords
Determinant: A numeric value that indicate singularity or
non-singularity of a square matrix.
E-
Identity: An identity is often represented by a triple hyphen.

Minor: A minor is the det. of the square matrix formed by deleting

one row and one column from some larger square matrix.

Cofactor: Cofactor is the multiplication of (–1)i + j with minor.

CC

Questions for Discussion

1. Find the value of the following determinant:
1 3 2
3 9 5
1 3 2

2. What are the main properties of determinants? Illustrate with

suitable examples.
3. By expanding the determinants, prove the identities:
a1 b1 a1 a2 a1 b1 a2 b1
(c)

(a) ≡ (b) ≡−
a2 b2 b1 b2 a2 b2 a1 b1

a1 b1 a + pa 2 b1 + pb2 a + qb1 b1
(c) ≡ 1 ≡ 1
a2 b2 a2 b2 a2 + qb2 b2
 Quantitative Techniques for Management Applications

102
4. Evaluate the determinants:

S
Notes
1 0 1 0 1 2 1 2
___________________ (a) (b) (c) (d)
0 1 1 1 1 4 3 4
___________________
3 1 −1 2 3 6 4 11
___________________ (e) (f) (g) (h)

E
2 4 −3 4 2 4 7 13
___________________
5. Evaluate the following determinants, first using property (9)
___________________
and then the other properties as appropriate:
___________________
3 0 2 −2 3 2

UP
___________________ (a) 7 0 5 (b) 0 5 0
___________________ 4 −1 2 3 −2 21
___________________
3 0 −6 2
9 21 11
___________________ −2 0 4 5
(c) 0 0 −3 (d)
−1 2 7 −3
5 15 −7
5 0 −10 17

2 10 −5 3
E-
4 −8 0 7
(e)
0 0 0 5
−3 6 1 9

6. Evaluate the following determinants, first using property (4)

three times, then property (3) once or twice to make the
number smaller and finally property (9), or other properties as
CC

appropriate.

2 3 1 −3 −44 30
(a) 4 6 1 (b) 2 5 −18
12 24 2 7 35 −63

1 3 1 2
32 35 3
6 12 4 12
(c) 28 10 4 (d)
1 6 3 8
34 40 3
4 9 2 2

2 3 6 4
(c)

3 −1 3 8
(e)
5 −3 3 4
12 −6 9 12
UNIT 7: Determinants

103
7. Four boys order in a fish-and-chips restaurant. A orders fish,

S
Notes
chips and coke. B orders two fish with chips. C orders fish and
coke. D orders chips and coke. The prices are ` 50 for fish, ` 18 ___________________
for chips, and ` 15 for coke. ___________________

(a) Express each boy’s order as a row vector.

E
___________________

(b) Add together these four vectors to obtain a fifth row vector ___________________
representing the total quantities ordered. ___________________
(c) Express the prices as a column vector. ___________________

UP
(d) Multiply each of the five row vectors by the price vector, to ___________________
obtain the amount owed by each boy and the total amount
___________________
owed.
___________________
(e) Check that the fifth result in (d) is equal to the sum of the
___________________
other four results.

Further Readings
E-
Books
R S Bhardwaj, Mathematics for Economics and Business, Excel
Books, New Delhi, 2005

D C Sanchethi and V K Kapoor, Business Mathematics

Sivayya and Sathya Rao, An Introduction to Business Mathematics

CC

Web Readings
www.textbooksonline.tn.nic.in

www.mathbusiness.com
(c)
 Quantitative Techniques for Management Applications

104

S
Notes

___________________

___________________

___________________

E
___________________

___________________

___________________

UP
___________________

___________________

___________________

___________________
E-
CC
(c)
UNIT 8: Probability

Unit 8
105

S
Activity
Notes
What does probability mean?

Probability
___________________

___________________

E
___________________
Objectives
___________________
After completion of this unit, the students will be aware of the following
topics: ___________________

\ Probability ___________________

UP
\ Concept of probability ___________________
\ Objective and subjective probability
___________________

___________________
Introduction
___________________
If all business decisions could be made under conditions of
certainty, the only valid justification for a poor decision would be
failure to consider all the pertinent facts. With certainty, one can
make a perfect forecast of the future. Unfortunately, however, the
E-
manager rarely, if ever, operates in a world of certainty. Usually,
the manager is forced to make decisions when there is uncertainty
as to what will happen after the decisions are made. In this latter
situation, the mathematical theory of probability furnishes a tool
that can be of great help to the decision maker.
CC

Concept of Probability
The idea of probability is normally associated and remembered in
connection with games or gambling. If a bookmaker offers odds of 2
to 1 against a team winning the world cup, it means that he
considers that the probability, that the team will win, is not more
than 1/3. When he accepts a bet of ` 1, he believes he has a
probability of at least 2/3 that he will gain the money and a
probability of not more than 1/3 that he will have to pay out ` 2 as
well as return the original ` 1. So he expects to make a profit, not
necessarily on this particular bet but on all the bets he takes when
(c)

applying the same general principles.

Herein lies the basis for a definition of probability as the ‘degree of
belief’ that an event will occur. Most statisticians would not accept
this as a satisfactory definition, but there is an important minority,
usually known as Bayesians, who do in fact take ‘degree of belief’
 Quantitative Techniques for Management Applications

106 as their definition of probability. The fact that there are differences

S
Notes of opinion even over the definition of probability will serve as a
___________________ warning to the reader of the difficulties of the subject, for there
were no such disagreements over the subjects covered in the
___________________
previous units.
___________________

E
It may be added that it is only the early quotations of betting odds
___________________
that are based on the bookmaker’s degree of belief. When a
___________________ reasonable number of bets have been taken, the bookmaker will
___________________ adjust the odds in such a way that he makes a profit whichever

UP
team wins. It is the balance of opinion among punters, which really
___________________
decide the odds.
___________________
An alternative definition of probability is based on counting
___________________
numbers of equally likely events. In a situation in which there are
___________________ ‘n’ possible outcomes, all equally likely, the probability of each of
these outcomes is 1 divided by ‘n’. The outcomes must be defined in
such a way that exactly one will occur; that is, it must be
impossible for none of them or for more than one of them to occur.
They are termed as simple events.
E-
If an ordinary six-sided die is thrown, the possible outcomes are
the numbers 1, 2, 3, 4, 5 and 6. One and only one of these will in
fact occur. If the die is unbiased, which is a way of saying that all
the outcomes are equally likely, then the probability of each simple
event is 1/6.
CC

It is now possible to discuss compound events, such as ‘less than 3’,

‘3 or more’, or ‘an even number’ when throwing an unbiased die. If
a compound even is defined to include m equally likely simple
events, then the probability of the compound event is m/n. The first
example given consists of the simple events 1, 2; the second
consists of 3, 4, 5, 6; and the third consists of 2, 4, 6. So their
respective probabilities are 1/3, 2/3, and 1/2.
This definition of probability is called the classical or a priories
definition of probability. It is ‘a priories’ because the probability
can be determined before the die has even been thrown. But it has
two great weaknesses. Its theoretical weakness is that the phrase
(c)

‘equally likely’ is synonymous with ‘equally probable’, and so the

definition is circular. Its practical weakness is that it cannot cope
with situations where the simple events are not all equally likely;
it would be an unusual horse-race in which one could say with
confidence that all the horses are equally likely to win.
UNIT 8: Probability

107
Later in this unit a third definition of probability will be

S
Notes
Activity
introduced. However, the classical definition is perfectly adequate
What are objective and
for discussing permutations and combinations, and nothing said ___________________
subjective probabilities?
later will detract from the validity of the conclusions reached using ___________________
the classical definition.

E
___________________

Check Your Progress ___________________

Fill in the blanks: ___________________

1. An alternative definition of probability is based on ___________________

UP
__________ numbers of equally likely events. ___________________

2. An ordinary die has _________ faces. ___________________

___________________
Objective and Subjective Probabilities ___________________

Most of us are familiar with the laws of chance regarding coin

flipping. If someone asks about the probability of a head on one
toss of a coin, the answer will be one-half, or 0.50. This answer is
based on common experience with coins, and assumes that the coin
E-
is a fair coin and that it is “fairly” tossed. This is an example of
objectivity probability. There are two interpretations of objective
probability. The first relies on the symmetry of outcomes and
implies that outcomes that are identical in essential aspects should
have the same probability. A fair coin is defined to be one that is
evenly balanced and has two sides that are identical (except for
CC

minor differences in the image). Hence, each side should have

equal probability of one-half (ignoring the possibility of the coin
landing on its edge). If the coin was bent, weighted, or two-tailed,
the answer would be different. As another example, suppose we
have a box containing three red and seven black balls (that are of
the same size, have the same feel, and are otherwise identical
except for colour), and the balls are thoroughly mixed. The
symmetry of outcome interpretation would assign a 0.10
probability to each ball, and hence a 0.30 chance of drawing a red
ball.
The relative frequency interpretation of objective probability relies
(c)

on historical experience in identical situations. Thus, if a coin has

been flipped 10,000 times with 4,998 heads, we would conclude
that the probability was 0.50 (i.e. 4,998/10,000 rounded) for a head
the next time the coin was flipped in the same manner as before.
 Quantitative Techniques for Management Applications

108
A subjective interpretation of probabilities is often useful for

S
Notes
Activity
business decision making. In the case of objective probability,
What is the basic statement of
___________________
probability? definitive historical information, common experience (objective
___________________ evidence) or rigorous analysis lies behind the probability
___________________ assignment. In the case of subjective interpretation, quantitative

E
historical information may not be available; and instead of
___________________
objective evidence, personal experience becomes the basis of the
___________________
probability assignment. For managerial decision-making
___________________ purposes, the subjective interpretation is frequently required, since

UP
___________________ reliable objective evidence may not be available.
___________________ Assume that a manager is trying to decide whether or not to build
___________________ a new factory, and the success of the factory depends largely on
whether or not there is a recession in the next five years. A
___________________
probability assigned to the occurrence of a recession would be a
subjective weight, which would be assigned after. There would
certainly be less agreement on this probability than there would be
on the probabilities of drawing a red ball, or of a fair coin coming
E-
up heads. Since we are primarily concerned in this book with
management decisions, we shall often assign subjective
probabilities to events that have a critical bearing on the
management decision. This procedure aims to ensure consistency
between a decision-maker’s judgment about the likelihood of the
possible states of nature and the decision that is made.
CC

One important objective of the suggested decision process is to

allow the decision maker to think in terms of the possible events
that may occur after a decision, the consequences of these events
and the probabilities of these events and consequences, rather
than having the manager jump immediately to the question of
whether or not the decision is desirable.

Check Your Progress

Fill in the blanks
1. A _________ interpretation of probabilities is often
(c)

useful for business decision making.

2. In the case of objective probability, definitive historical
information, common experience (objective evidence), or
rigorous analysis lie behind the _________.
UNIT 8: Probability

Basic Statement of Probability 109

S
Notes
Activity
Two fundamental statements about probabilities are: What are mutually exclusive
___________________
events?
1. Probabilities of all the various possible outcomes of a trial ___________________
must sum to one.

E
___________________
2. Probabilities are always greater than or equal to zero (i.e.,
___________________
probabilities are never negative) and are less than or equal to
one. The smaller the probability, the less likely the chance of ___________________

the event happening. ___________________

UP
The first statement indicates that if A and B are the only ___________________
candidates for an office, the probability that A will win plus the ___________________
probability that B will win must sum to one (assuming a tie is not
___________________
possible).
___________________
The second statement results in the following interpretations. If
an event has a positive probability, it may possibly occur; the event
may be impossible, in which case it has a zero probability; or the
event may be certain to occur, in which case the probability is
E-
equal to one. Regardless of whether probabilities are interpreted as
objective probabilities or as subjective weights, it is useful to think
in terms of a weighting scale running from zero to one. If someone
tosses a coin of unknown characteristics 500 times to obtain an
estimate of objective probabilities and the results are 225 heads
and 275 tails, the range of possible results may be converted to a
zero-to-one scale by dividing by 500. The actual results are
CC

225/500 = 0.45 heads and 275/500 = 0.55 tails. Hence, if we wish to

derive probabilities, we shall manipulate the data so as to adhere
to the zero-to-one scale. The 0.45 and the 0.55 may be used as the
estimators of the true probabilities of heads and tails (the true
probabilities are unknown).

Check Your Progress

Fill in the blanks:
1. Probabilities of all the various possible outcomes of a
trial must sum to _________.
(c)

2. Probabilities are never _________ .

Mutually Exclusive Events

Two or more events are mutually exclusive if only one of the events
can occur on any one trial. The probabilities of mutually exclusive
 Quantitative Techniques for Management Applications

110
events can be added to obtain the probability that one of a given

S
Notes collection of the events will occur.
___________________
Example 8.1
___________________
The probabilities shown in Table 8.1 reflect the subjective estimate
___________________
of a newspaper editor regarding the relative chances of four

E
___________________ candidates for a public office (assume a tie is not possible).
___________________ Table 8.1: Election Probabilities

___________________ Event: Elect Probability

UP
___________________ Candidate A 0.18
Candidate B 0.42
___________________
Candidate C 0.26
___________________
Candidate D 0.14
___________________ 1.00

These events are mutually exclusive, since in one election (or in

one trial) only one event may occur; therefore, the probabilities are
additive. The probability of a Democratic victory is 0.60; of a
E-
Republican victory, 0.40; or of either B or C winning, 0.68. The
probability of both B and C winning is zero, since only one of the
mutually exclusive events can occur on any one trial.

Check Your Progress

Fill in the blanks:
CC

1. Two or more events are mutually exclusive if __________

of the events can occur on any one trial.
2. The probability of mutually exclusive events can be
added to obtain to probability that one of a given
_______ of the events will occur.

Dependent and Independent Events

Events may be either independent or dependent. If two events are
(statistically) independent, the occurrence of one event will not
affect the probability of the occurrence of the second event.
(c)

When two (or more) events are independent, the probability of both
events (or more than two events) occurring is equal to the product
of the probabilities of the individual events. That is:

P (A and B) = P(A) × P(B) if A, B independent

UNIT 8: Probability

111
Where

S
Notes
P (A and B) = Probability of events A and B both occurring ___________________

P (A) = Probability of event A occurring ___________________

P (B) = Probability of event B occurring

E
___________________

___________________
The above equation indicates that the probability of A and B both
occurring is equal to the probability of A multiplied by the ___________________

probability of B, if A and B are independent. If A is the probability ___________________

UP
of a head on the first toss of the coin and B is the probability of a
___________________
head on the second toss of the coin, then:
___________________
P (A) =½
___________________
P (B) =½ ___________________

P (A and B) =½x ½=¼

The probability of A and B occurring (two heads) is one-fourth. P

(A and B) is the joint probability of events A and B. Where
E-
appropriate, the word and can be omitted to simplify the notation
and the joint probability can be written simply as P (AB).

To define independence mathematically, we need the symbol P

(B½A). The symbol P (B½A) is read “the probability of event B,
given that event A has occurred.” P (B½A) is the conditional
probability of event B, given that event A has taken place. Note
CC

that PB½A) does not mean the probability of event B divided by A

– the vertical line followed by A means “given that event A has
occurred.”

With independent events:

P (B½A) = P (B) if A, B independent

That is, the probability of event B, given that event A has occurred,
is equal to the probability of event B if the two events are
independent. With two independent events, the occurrence of the
one event does not affect the probability of the occurrence of the
(c)

second [in like manner, P (A½B) =P (A)].

Two events are dependent if the occurrence of one of the events

affects the probability of the occurrence of the second event.

Let’s take an example.

 Quantitative Techniques for Management Applications

112
Flip a fair coin and determine whether the result is heads or tails.

S
Notes
If heads, flip the same coin again. If tails, flip an unfair coin that
___________________
has a three-fourths probability of heads and a one-fourth
___________________
probability of tails. Is the probability of heads on the second toss in
___________________ any way affected by the results of the first toss? The answer here is

E
___________________ yes, since the result of the first toss affects which coin (fair or
___________________ unfair) is to be tossed the second time.

___________________ Another example of dependent events involves mutually exclusive

UP
___________________ events. If events A and B are mutually exclusive, they are
dependent. Given that event A has occurred, the conditional
___________________
probability of B occurring must be zero, since the two events are
___________________
mutually exclusive. Let’s take up two examples where we clear the
___________________ difference between joint probabilities, conditional probabilities and
unconditional (or marginal) probabilities.

Example 8.2

Assume we have three boxes, which contain red and black balls as
E-
follows:

Box 1 : 3 red and 7 black

Box 2 : 6 red and 4 black

Box 3 : 8 red and 2 black

CC

Suppose we draw from a ball from box 1; if it is red, we draw a ball

from box 2. If the ball drawn from box 1 is black, we draw a ball
from box 3. Consider the following probability questions about this
game:

1. What is the probability of drawing a red ball from box 1? This

probability is an unconditional or marginal probability; it is
0.30. (The marginal probability of getting a black is 0.70).

2. Suppose we draw a ball from box 1, and it is red; what is the

probability of another red ball when we draw from box 2 on the
(c)

second draw? The answer is 0.60. This is an example of a

conditional probability. That is, the probability of a red ball on
the second draw if the draw from box 1 is red is a conditional
probability.
UNIT 8: Probability

113
3. Suppose our first draw from box 1 was black; then the

S
Notes
conditional probability is 0.80. The draw from box 1 (the
___________________
conditioning event) is very important in determining the
probabilities of red (or black) on the second draw. ___________________

E
___________________
4. Suppose, before we draw any balls, we ask the question: What
is the probability of drawing two red balls? This would be a ___________________

joint probability; the event would be a red ball on both draws. ___________________
The computation of this joint probability is a little more ___________________

UP
complicated than the above questions, and some analysis will
___________________
be of value. Computations are as follows:
___________________

___________________

___________________
E-
Figure 8.1
CC

Table 8.2 shows the joint probability of two red balls as 0.18 [i.e.
P(R and R) or more simply P(RR), the top branch of the tree]. The
joint probabilities may be summarized as follows:
Two red balls P(RR) = 0.18
A red ball on first draw and a black ball on second draw P(RB) = 0.12
A black ball on first draw and a red ball on second draw P(BR) = 0.56
Two black balls P(BB) = 0.14
1.00

Table 8.2: Probabilities Calculations

Marginal z Conditional = Joint

Event P(A) P(B½A) = P(A and B)
(c)

RR P(R) = 0.30 P(R|R) = 0.60 P(RR) = 0.18

RB P(R) = 0.30 P(B|R) = 0.40 P(RB) = 0.12
BR P(B) = 0.70 P(R|B) = 0.80 P(BR) = 0.56
BB P(B) = 0.70 P(B|B) = 0.20 P(BB) = 0.14
 Quantitative Techniques for Management Applications

114
Check Your Progress

S
Notes
Activity
What is the importance of Fill in the blanks:
___________________
decision tree ?
___________________
1. With two independent events, the occurrence of the one
event ___________ affect the probability of the
___________________

E
occurrence of the second.
___________________
2. The events are dependent if the occurrence of one of the
___________________ events ___________ two probability of the second event
___________________ occurrence.

UP
___________________

___________________ Decision Trees

___________________ This is a very useful device for illustrating uncertain situations. The
___________________ first fork shows that either a red or a black may be drawn, and the
probabilities of these events are given. If a red is drawn, we go to box
2, where again a red or black may be drawn, but with probabilities
determined by the fact that the draw will take place in box 2. For the
second forks, we have conditional probabilities (the probabilities
E-
depend on whether a red or a black ball was chosen on the first draw).
At the end of each path are the joint probabilities of following that
path. The joint probabilities are obtained by multiplying the
marginal (unconditional) probabilities of the first branch by the
conditional probabilities of the second branch.
Table 8.3 presents these results in a joint probability table; the
CC

intersection of the rows and columns are joint probabilities. The

column on the right gives the unconditional probabilities
(marginals) of the outcome of the first draw; the bottom row gives
the unconditional or marginal probabilities of the outcomes of the
second draw. Table 8.3 effectively summarizes the tree diagram.
Now, let us compute some additional probabilities:
1. Probability of one red and one black ball, regardless of order:
= 0.56 + 0.12 = 0.68
2. Probability of a black ball on draw 2:
Explanatory calculation:
(c)

Probability of red-black = 0.12

Probability of black-black = 0.14
Probability of black on draw 2 = 0.26
3. Probability of second draw being red if first draw is red:
= 0.60
UNIT 8: Probability

115
If first draw is red, we are in the R row of

S
Notes
Table 8.3, which totals 0.30. The question is
___________________
what proportion is 0.18 of 0.30?
___________________
The answer is 0.60; or in terms of the appropriate formula:

E
___________________
P(R 2 and R1 ) 0.18
P(R 2|R1 ) = = = 0.60 ___________________
P(R1 ) 0.30
___________________
Table 8.3: Joint Probability Table
___________________

UP
___________________

___________________

___________________

___________________
E-
Example 8.3
We give a further example of the basic probability definitions. A
survey is taken of 100 families; information is obtained about
family income and about whether the family purchases a speciality
food product. The results are shown in Table 8.4.

Table 8.4: Survey of 100 Families, Classified by Income and Buying

CC

Behaviour
Low Income High Income Total
(family income (family income of number of
below ` 30,000) ` 30,000 or more) Families
Family is:
Buyer of speciality 18 20 38
food products
Non-buyer 42 20 62
Total number of families 60 40 100

Suppose a family is to be selected at random from this group.

1. What is the probability that the family selected will be a
buyer? Since 38 of the 100 families overall are buyers, the
(c)

probability is 0.38. Note that this is a marginal or

unconditional probability.
2. What is the probability that the selected family is both a buyer
and with high income? Note that this is a joint probability.
P(Buyer and High income) = 20/100 = 0.20.
 Quantitative Techniques for Management Applications

116 3. Suppose that a family is selected at random and you are

S
Notes
Activity informed that it has high income. What is the probability that
What is combination means?
___________________ this family is a buyer? Note that this asks for the conditional
probability P (Buyer|High income). Of the 40 families with High
___________________
income, 20 are Buyers. Hence, the probability is 20/40 = 0.50.
___________________

E
4. Are the events Buyer and High income independent for this
___________________
group of families? Note from question 1 that P(Buyer) = 0.38;
___________________ from question 3, P(Buyer|High income) = 0.50. These are not
___________________ the same. Hence, the two events are dependent. Knowing the

UP
family has high income affects the probability that it is a
___________________
buyer. Another way of expressing this dependence is to say
___________________
that the percentage of buyer is not the same for the high – and
___________________ low-income families.
___________________
Revision of Probabilities
Having discussed joint and conditional probabilities, let us
investigate how probabilities are revised to take account of new
information.
E-
Example 8.4
Suppose we do not know whether a particular coin is fair or unfair. If
the coin is fair, the probability of a tail is 0.50; but if the coin is unfair,
the probability of a tail is 0.10. Assume we assign a prior probability
to the coin being fair of 0.80 and a probability of 0.20 to the coin being
CC

unfair. The event “fair coin” will be designated A1, and the event
“unfair coin” will be designated A2. We toss the coin once; say, a tail
is the result. What is the probability that the coin is fair?
The conditional probability of a tail, given that the coin is fair, is
0.50; that is P(tail|A1) = 0.50. If the coin is unfair, the probability
of a tail is 0.10; P(tail|A2) = 0.10

Let us compute the joint probability P (tail and A1). There is an

initial 0.80 probability that A1 is the true state; and if A1 is the true
state, there is a 0.50 conditional probability that a tail will result.
The joint probability of state A1 being true and obtaining a tail is
(c)

(0.80 x 0.50) = 0.40. Thus:

P(tail and A1) = P(A1) × P(tail|A1) = 0.80 x 0.50 = 0.40

The joint probability of a tail and A2 is equal to:

P(tail and A2) = P(A2) × P(tail|A2) = 0.20 × 0.10 = 0.02

UNIT 8: Probability

117
A tail can occur in combination with the state “fair coin” or in

S
Notes
combination with the state “unfair coin”. The probability of the
former combination is 0.40; of the latter, 0.02. The sum of the ___________________
probabilities gives the unconditional probability of a tail on the ___________________
first toss; that is, P(tail) = 0.40 + 0.02 = 0.42:

E
___________________
P(tail and A2) = 0.02 ___________________
P(tail and A1) = 0.40 ___________________
P(tail) = 0.42 ___________________

UP
If a tail occurs, and if we do not know the true state, the ___________________
conditional probability of state A1 being the true state is:
___________________
P(tail and A1 ) 0.40
P(A1|tail) = = = 0.95 ___________________
P(tail) 0.42
___________________
Thus, 0.95 is the revised or posterior probability of A1, given that a
tail has occurred on the first toss.
Similarly:
P(tail and A 2 ) 0.02
P(A1|tail) = = = 0.05
E-
P(tail) 0.42
In more general symbols:
P(A i and B)
P(A i|B) =
P(B)
Conditional probability expressed in this form is known as Bayes
theorem. It has many important applications in evaluating the
CC

worth of additional information in decision problems.

In this example, the revised probabilities for the coin are 0.95 that
it is fair and 0.05 that it is unfair (the probabilities were initially
0.80 and 0.20). These revised probabilities exist after one toss
when the toss results in a tail. It is reasonable that the probability
that the coin is unfair has decreased, since a tail appeared on the
first toss, and the unfair coin has only a 0.10 probability of a tail.

Check Your Progress

Fill in the blanks:
(c)

1. Decision trees is a very useful device for illustrating

_____________ situations.
2. The joint probabilities are obtained by ________ the
marginal probabilities of the first branch by conditional
probabilities of the second branch.
 Quantitative Techniques for Management Applications

118
Combinations

S
Notes
Activity
What is the importance of It is sometimes necessary to know the number of different ways in
___________________
decision tree ? which ‘r’ objects can be selected from n objects without regard to
___________________ sequence. For instance, the number of different permutations of
___________________ given cards from a pack of 52 cards is 52P5. The same five cards

E
___________________ dealt in different sequences are different permutations. In practice,
the interest is usually in the number of different possible hands
___________________
irrespective of the sequence in which they were dealt. To obtain
___________________
this, one must divide by the number of ways in which five cards

UP
___________________ can be arranged among themselves, which is 5P5.
___________________ Each different way of selecting r objects from n without regard to
___________________ the sequence in which they were selected is termed a combination,
___________________
and each such combination consists of rPr permutations. The
number of combinations is represented by nCr, and so nCr divided by
Pr, since rPr is r! :
r

n!
n
Cr =
(n − r)!r !
E-
It will be seen that nCr works out the same as nCn-r. This is
reasonable, since the number of ways of selecting five cards from a
pack of 52 is obviously the same as the number of ways of selecting
47 cards and leaving a hand of five cards behind.

Example 8.5
CC

In how many different ways three bolts can be selected from a box
containing eight bolts?

Solution:
The answer is 8C3. It is convenient to adopt the practice of using
dots in place of multiplication signs when several numbers are all
multiplied together:
8! 8.7.6
8
C3 = = = 56
5!3! 1.2.3
There is no need to write out the factorials in full. The number of
(c)

integers to be multiplied in both numerator and denominator is the

smaller of r and (n-r). The denominator always cancels completely,
since nCr must be an integer.
UNIT 8: Probability

119
Example 8.6

S
Notes
If three bolts are selected at random from a box containing six
___________________
sound and two faulty bolts, what is the probability of obtaining (i)
three sound, (ii) two sound and one faulty, (iii) one sound and two ___________________
faulty bolts?

E
___________________

Solution: ___________________

The word ‘random’ indicates that all the 56 different ways of ___________________
selecting three bolts from a total of eight bolts are equally likely, ___________________

UP
and so the classical definition of probability can be applied. The
___________________
number of ways of obtaining three sound bolts is 6C3, which is 20,
___________________
and dividing this by 56 gives 0.357.
___________________
The number of ways of obtaining two sound bolts is 6C2, which is
15. Each of these combinations can be associated with either of the ___________________
2
C1 ways of obtaining one faulty bolt, and so 15 is multiplied by two
and divided by 56 to give 0.536.
Following the same principles, the probability of obtaining one
E-
sound and two faulty bolts is:
6
C1 2C2 6 × 1 3
= = = 0.107
8
C3 56 28

A check confirms that the three probabilities calculated of possible

combinations of sound and faulty bolts total 1, which must be so
CC

since one of the three results must occur.

Probabilities may be similarly calculated in any situation involving
permutations or combinations provided that, all the outcomes are
equally likely. Using example 8.4, the probability of choosing the
winning list from ten models in a fashion competition is 1 divided
by 1,51,200 if six models have to be listed, provided that no skill is
involved.

Check Your Progress

Fill in the blanks
1. There is _________need to write out the factorials in full.
(c)

2. Probabilities may be calculated in any situation

involving permutations or combinations provided that,
all the outcomes are __________likely
 Quantitative Techniques for Management Applications

120
Summary

S
Notes
The probabilities assigned to various events on the basis of the
___________________
conditions of the experiment or by actual experimentation or past
___________________ experience or on the basis of personal judgement are called prior
___________________ probabilities.

E
___________________

___________________ Lesson End Activity

___________________ A and B stand in a ring with 10 other persons. If arrangement of

UP
12 persons is at random, find the chance that there are exactly
___________________
three persons between A and B.
___________________

___________________
Keywords
___________________
A Sample Event: The basic possible outcome of an experiment, it
cannot be broken down into simpler outcomes.
Event: Any set of all possible outcomes or simple events of an
experiment.
E-
Probability: A numerical measures of the likelihood of occurrence
of an uncertain event.
Conditional Probability: The probability of an event occurring,
given that another event has occurred.

Questions for Discussion

CC

1. Define the term 'probability' by (a) The Classical Approach, (b)

The Statistical Approach. What are the main limitations of
these approaches?
2. Discuss the axiomatic approach to probability. In what way it
is an improvement over classical and statistical approaches?
3. What do you mean by objective and subjective probability?
4. Discuss the basic statements of probability.
5. Distinguish between objective probability and subjective
probability. Give one example of each concept.
(c)

6. State and prove theorem of addition of probabilities for two

events when (a) they are not independent, (b) they are
independent.
UNIT 8: Probability

Further Readings 121

S
Notes
Books ___________________

R S Bhardwaj, Mathematics for Economics and Business, Excel ___________________

Books, New Delhi, 2005

E
___________________

D C Sanchethi and V K Kapoor, Business Mathematics ___________________

Sivayya and Sathya Rao, An Introduction to Business Mathematics ___________________

___________________
Web Readings

UP
___________________
www.managementstudyguide.com
___________________
www.textbooksonline.tn.nic.in
___________________
www.mathbusiness.com ___________________
E-
CC
(c)
 Quantitative Techniques for Management Applications

122

S
Notes

___________________

___________________

___________________

E
___________________

___________________

___________________

UP
___________________

___________________

___________________

___________________
E-
CC
(c)
UNIT 9: Random Variables and Probability Distributions

Unit 9
123

S
Activity
Notes
What is probability

Random Variables and Probability

___________________
distribution?
___________________

Distributions

E
___________________

___________________

Objectives ___________________
After completion of this unit, the students will be aware of the following
___________________

UP
topics:
___________________
\ Random Variables
\ Binomial distribution ___________________

\ Poisson distribution ___________________

___________________

Introduction
A probability function is a rule that assigns probabilities to each
element of a set of events that may occur. In this unit, we will
E-
discuss random variables and probablity distribution.

Random Variables
We can assign a specific numerical value to each element of the set
of events, a function that assigns these numerical values is termed
a random variable. The value of a random variable is the general
CC

outcome of a random (or probability) experiment. It is useful to

distinguish between the random variable itself and the values that
it can take on. The value of a random variable is unknown until
the event occurs (i.e., until the random experiment has been
performed). However, the probability that the random variable
will be any specific value is known in advance.
The probability of each value of the random variable is equal to the
sum of the probabilities of the events assigned to that value of the
random variable.
For example, suppose we define the random variable Z to be the
(c)

number of heads in two tosses of a fair coin. Then the possible

values of Z, and the corresponding probabilities, are:
 Quantitative Techniques for Management Applications

124
Possible Values of Z Probability of Each Value

S
Notes
0 ¼
___________________
1 ½
___________________
2 ¼
___________________

E
Random variables can be grouped into probability distribution,
___________________ which can be either discrete or continuous. Discrete probability
___________________ distributions are those in which the random variable can take on
only specific values. The table above is an example of such a
___________________
distribution since the random variable Z can be only 0, 1, or 2. A

UP
___________________
continuous probability distribution is one in which the value of the
___________________ random variable can be any number within some given range of
___________________ values – say, between zero and infinity. For example, if the random
variable was the height of members of a population, a person could
___________________
be 5.3 feet, 5.324 feet, 5.32431 feet, and so on, depending on the
ability of instruments to measure. Some additional examples of
random variables are shown in the Table 9.1.
A discrete probability distribution is sometimes called a probability
E-
mass function (p.m.f.) and a continuous one is called a probability
density function (p.d.f.) Graphs of the two types of distributions are
shown in Figure 9.1 and 9.2.
For a discrete distribution, the height of each line represents the
probability for that value of the random variable. For example, 0.30
is the probability that tomorrow’s demand will be 0.2 tons in figure.
CC

For a continuous random variable, the height of the probability

density function is not the probability for an event. Rather, the area
under the curve over any interval on the horizontal axis represents
the probability of taking on a value in that interval.

Table 9.1: Examples of Random Variables

(c)
UNIT 9: Random Variables and Probability Distributions

125

S
Activity
Notes
What is probability mass
___________________
function?
What___________________
is continuous probability
distribution?

E
___________________

___________________

___________________

___________________

UP
___________________

___________________

___________________

___________________
Figure 9.1: Discrete Probability Distribution
(Probability Mass Function or p.m.f.)

Now let us take a look at the first discrete probability distribution,

the binomial probability distribution and how it is derived. The
E-
binomial probability distribution is the base on which several other
probability distributions are based upon.
CC

Figure 9.2: Continuous Probability Distribution

Check Your Progress

Fill in the blanks:
1. The binomial probability distribution is the base on
(c)

which several other probability distributions are

_______________.
2. ___________ variables can be grouped into probability
distribution, which can be either discrete or continuous.
 Quantitative Techniques for Management Applications

126
The Bernoulli Process

S
Notes
Activity
What is Bernoulli?
___________________ Understanding of the Bernoulli process is necessary before we
understand the binomial distribution. The characteristics of a
___________________
Bernoulli process are described below:
___________________

E
1. The outcomes or results of each trial in the process are
___________________
characterized as one of two types of possible outcomes, such
___________________ as:
___________________ a. Success, failure.

UP
___________________
b. Yes, no.
___________________
c. Heads, tails.
___________________
d. Zero, one.
___________________
2. The probability of the outcome of any trial is “stable” and does
not change throughout the process. For example, the
probability of heads, given a fair coin, is 0.50 and does not
change, regardless of the number of times the coin is tossed.
E-
3. The outcome of any trial is independent of the outcome of any
previous trial. In other words, the past history of the process
would not change the probability assigned to the next trial. In
our coin example, we would assign a probability of 0.50 to the
next toss coming up heads; even if we had recorded heads on
the last 10 trials (we assume the coin is fair).
CC

4. The number of trials is discrete and can be represented by an

integer such as 1, 2, 3, and so on.
Given a process, we may know that it is Bernoulli, but we may or
may not know the stable probability characteristic of the process.
With a fair coin, we may know the process is Bernoulli, with
probability 0.50 of a success (say heads) and probability 0.50 of a
failure (tails). However, if we are given a coin and told it is not
fair, the process (flipping the coin) may still be Bernoulli, but we do
not know the probability characteristic. Hence, we may have a
Bernoulli process with a known or unknown probability
characteristic.
(c)

Many business processes can be characterized as Bernoulli for

analytical purposes, even though they are not true Bernoulli in
every respect. If the “fit” is close enough, we may assume that the
Bernoulli process is a reasonable characterization. Let us discuss
some examples.
UNIT 9: Random Variables and Probability Distributions

127
Suppose we are concerned with a production process where a

S
Notes
certain part (or product) is produced on a machine, we may be
interested in classifying the parts as “good” or “defective”, in which ___________________
case the process may be Bernoulli. If the machine is not subject to
___________________
fast wear and tear – that is, if a setting will last for a long run of
parts 0 the probability of good parts may be sufficiently stable for

E
___________________
the process to qualify as Bernoulli. If, on the other hand, more ___________________
defects occur as the end of the run approaches, the process is not
___________________
Bernoulli. In many such processes, the occurrence of good and
defective parts is sufficiently stable (no pattern over time is ___________________

UP
observable) to call the process Bernoulli. The probability of good
___________________
and defective parts may remain stable through a production run,
but it may vary from run to run (because of machine setting, for ___________________
example). Here, the process could still be considered Bernoulli, but ___________________
the probability of a success (or failure) will change from run to run.
___________________
A different example of a Bernoulli process is a survey to determine
whether or not consumers prefer liquid to powdered soaps. The
outcome of a survey interview could be characterized a “yes”
(success) or “no” (failure) answers to the question. If the sample of
consumers was sufficiently randomized (no pattern to the way in
E-
which the yes or no answers occur), Bernoulli (with an unknown
probability) may be a useful description of the process.
Note that if the probability of a success in a Bernoulli process is
0.50, the probability of a failure is also 0.50 (since the probabilities
of the event happening and the event not happening add to one). If
the probability of a success is p, the probability of a failure is (1 – p).
CC

Check Your Progress

Fill in the blanks:
1. A different example of a Bernoulli process is a
__________ to determine whether or not consumers
prefer liquid to powdered soaps.
2. Many business processes can be characterized as
Bernoulli for _______ purposes.

The Binomial Theorem

(c)

Let us start by using an example.

Example 9.1

The probability that a salesman makes a sale on a visit to a

prospect is 0.2.
 Quantitative Techniques for Management Applications

128
What is the probability, in 2 visits, of:

S
Notes
z making no sales
___________________
z making one sale
___________________
z making two sales
___________________

E
___________________ Solution:

___________________ Here p = probability of sale = 0.2

___________________ And q = probability of no sale = 1 – 0.2 = 0.8

UP
___________________ The various outcome possibilities are
___________________ Visit 1 Visit 2 Probabilities
___________________ Sale Sale i.e. p x p = p2 = 0.22 = 0.04
___________________ Sale No sale i.e. p x q = 0.02 x 0.8 = 0.16

No sale Sale i.e. q x p = 0.8 x 0.2 = 0.16

No sale No sale i.e. q x q = q2 = 0.82 = 0.64

1.00
E-
Thus P (no sales) = 0.64
P (one sale) = 0.32
P (two sales) = 0.04
In this simple example, it is easy to show the whole process but
this becomes lengthy and cumbersome where the number of trials
(visits, in the above example) becomes larger.
CC

Fortunately there is a simpler approach which is by the expansion

of the binomial expression. The general form of the binomial
expression is
(p + q)n
Where p = probability of an event occurring
q = probability of an event not occurring
and n = number of trials
In the above example p = probability of a sale, i.e., 0.2,
q = probability of no sale, i.e., 0.8 and n = number of visits i.e. 2.
(c)

Use the binomial expansion to confirm the probabilities.

∴ (p + q)2 = p2 + 2pq + q2
= 0.22 + 2(0.2 x 0.8) + 0.82
= 0.04+0.32 + 0.64, i.e., the values previously obtained.
UNIT 9: Random Variables and Probability Distributions

129
Where n becomes larger it is useful to be able to calculate the

S
Notes
coefficients of each part of the expansion in a direct manner rather
than writing out the whole expansion. ___________________

It is easy to show by direct multiplication that: ___________________

E
___________________
(a + b)2 = a2 + 2ab + b2
___________________
(a + b)3 = a3 + 3a2b + 3ab2 + b3
___________________
These are called expansions of the expressions on the left, which
are called binomial expressions, because they each contain two ___________________

UP
terms within the bracket. The binomial theorem is concerned with ___________________
the general form of the expansion of (a+b).
___________________
It is not difficult to see from the above examples that the first term ___________________
in the expansion of (a + b)n is an, the last term is bn, and there are
___________________
(n + 1) terms in all. In general, the jth term consists of an+1–j bj-1
preceded by an appropriate coefficient, where j takes all integral
values from 1 to (n + 1).
To find the appropriate coefficient, one can investigate why the
E-
coefficient of a2 b2 in the expansion of (a + b)4 works out as 6.
Writing out the four terms, which have to be multiplied together:
(a + b) (a + b) (a + b) (a + b)
It is easy to see that multiplying a in the first bracket by a in the
second, b in the third, and b in the fourth will give the term a2b2.
But the same result is obtained by taking the sequence a, b, a, b,
CC

from the four brackets, or by taking any of the other possible

sequences which include ‘a’ from any two of the brackets and ‘b’
from the remaining two brackets.
The number of different ways of selecting two brackets from four is
4
C2 is six. The brackets selected are the ones which contribute the
symbol a, whilst the remaining brackets contribute the symbol b.
We can now consider the general term in the expansion of (a + b)n.
It is usual to refer to the jth term as the (r + 1)th term, so that the (r
+ 1)th term has the symbol an-r br and r takes all integer values from
0 to n. The coefficient in the (r + 1)th term is then the number of
(c)

ways of selecting (n-r) brackets from n, which is nCn –r.

It would be equally correct to select the r brackets which are to

provide the symbol b, letting the remaining (n - r) brackets provide
the symbol a. The coefficient is then nCr, but it has already been
 Quantitative Techniques for Management Applications

130 seen earlier that nCr is always equal to nCn–r and so there is no

S
Notes
contradiction. It follows that the series of coefficients is always
___________________ symmetrical; the coefficient of a n–r br is always the same as the
___________________ coefficient of ar bn-r.

___________________ Hence the binomial theorem states that for any positive integer n:

E
___________________ (a + b)n = an + nC1an-1b + nC2an-2b2 +....+ nCran–rbr +....+nCn-1abn-1 + bn
___________________ Sometimes nC1 and nCn-1 are written simply as n. On the other
___________________ hand, one can complete the uniformity of the expansion by writing

UP
___________________ the coefficients in the first and last terms as nC0 , nCn respectively,

___________________
these both being equal to 1.

___________________ Example 9.2

___________________ For components of a certain type, the probability that a component
is faulty when it leaves the production line is 0.05. If 10
components are selected at random, what is the probability of
obtaining two faulty components?

Solution:
E-
This is the same problem as in example 9.2 except that the size of
the batch must now be regarded as infinite. It is not difficult to see
that the previous approximate formula now becomes exact, so that
the answer is 0.0746.
The exact probability distribution for the number of faulty
CC

components in example 9.2 is called the hyper geometric

distribution. It is not very important in practice. If a sample is
drawn from a large batch, the binomial distribution is usually a
good enough approximation. If only a small batch is involved, it
can be examined in its entirety and then no probability problem
arises.
In some type of problems the binomial distribution can be derived
directly from a consideration of probabilities without reference to
batch sizes.

Example 9.3
(c)

The probability of meeting the buyer at a random visit to a certain

firm is 0.05. If a salesman makes 10 random visits, what is the
probability that he will meet the buyer on two occasions?
UNIT 9: Random Variables and Probability Distributions

Solution: 131

S
Notes
The probability that the salesman will meet the buyer on both the
first two visits is 0.05 multiplied by 0.05. The probability law of ___________________
multiplication states that the probability that both of two events ___________________
will occur is the product of their separate probabilities, provided

E
___________________
that the events are independent. Two events are said to be
independent if the fact that one has occurred makes no difference ___________________

to the probability that the second will occur. We can assume that ___________________
this will apply to random visits by the salesman.
___________________

UP
Further visits are again independent. The probability of not ___________________
meeting the buyer is 0.95. Multiplying all the probabilities, it
___________________
follows that the probability of meeting the buyer on the first two
occasions and not on any of the following eight occasions is: ___________________

(0.05)2 (0.95)8 ___________________

The probability of meeting the buyer on none of the first eight

visits but on both of the last two visits is similarly found to be:
(095)8 (0.05)2
E-
This is the same as the previous probability, and in fact any
arrangement of two successes among the 10 visits will have the
same probability.
The probability law of addition states that, the probability that
either of the two events will occur is the sum of their separate
probabilities, provided that they are mutually exclusive; that is,
CC

provided that it is impossible for both to occur. This applies to any

two different arrangements of two successes among 10 visits, and
so adding the probabilities for the 10C2 different arrangements
gives the total probability of two successes in 10 visits as:
10
C2(0.95)8 (0.05)2

which is again 0.0746

In this example the concept of probability was not derived from
counting equally likely events. The figure 0.05 indicates that on an
average one will meet the buyer five times in every 100 visits. This
(c)

does not mean that there will be exactly five successes in each 100
visits, but that the relative frequency of successes will tend to the
figure 0.05 in an unlimited number of visits. So probability is here
defined as the limiting relative frequency of a success; this is the
definition of probability most widely applied by statisticians.
 Quantitative Techniques for Management Applications

132
Check Your Progress

S
Notes
Fill in the blanks:
___________________

___________________
1. The concept of probability was not derived from
counting ___________ events.
___________________

E
2. Probability is also defined as the limiting relative
___________________
frequency of a ___________.
___________________

___________________ Probability Distributions

UP
___________________
A probability distribution is a rule that assigns a probability to
___________________ every possible outcome of an experiment. An event whose
___________________ numerical value is determined by the outcome of an experiment is
called a variate or often a random variable.
___________________
There are two kinds of probability distributions, discrete and
continuous. Discrete probability distributions are those where only
a finite number of outcomes are possible. For example, the throw of
a dice has a discrete probability distribution as only six outcomes
E-
are possible and are known beforehand. Continuous probability
distributions are those which represent continuously variable
random variables.

The Binomial Probability Distribution

The Binomial Probability Function
CC

If the assumptions of the Bernoulli process are satisfied and if the

probability of a success on one trial is p, then the probability
distribution of the number of successes, r, in n trials, is a binomial
distribution.

Binomial Probabilities on Spreadsheets

The Excel and Quattro spreadsheet programs have a function that
can be used for evaluating binomial probabilities, both individual
terms and cumulative probabilities. The form of the function is:
= BINOMDIST(r, n, p, 0 or 1)
(c)

Where r is the number of successes, n is the number of trials, and p

is the probability of success on each trial. The last term is either a
zero or one; if a zero is entered, the individual binomial term is
given; if a one is used, the cumulative value (of the £ type) is given.
Suppose, for example, we want the probability of exactly three
UNIT 9: Random Variables and Probability Distributions

successes in five trials, with a probability of success of p = 0.4 - P(r 133

S
= 3½n = 5, p = 0.4). This is: Notes

= BINOMDIST(3,5,0.4,0), which equals 0.2304 ___________________

For the probability of 3 or fewer successes in five trials, with p = ___________________

0.4 - P(r £ 3½n = 5, p = 0.4)

E
___________________

= BINOMDIST (3, 5, 0.4, 1), which equals 0.9130 ___________________

___________________
Example 9.4
Suppose we plan to toss a fair coin three times and would like to ___________________

UP
compute the following probabilities: ___________________

a. The probability of three heads in three tosses. ___________________

___________________
b. The probability of two or more heads in three tosses.
___________________
c. The probability of fewer than two heads in three tosses.
Possible Outcomes

HHH
HHT
E-
HTH
THH
TTH
THT
HTT
TTT
CC

In this example, the Bernoulli process p is 0.50, and a head

constitutes a success. The number of trials (n) is three.
The first probability is the probability of three heads (successes) in
three tosses (three trials), given that the probability of a head on
any one toss is 0.50. This probability can be written as follows:
P(r = 3½p = 0.50, n = 3) =?
Where P = Probability, r = Number of successes, n = Number of
trials, and p = Probability of success on any one trial. The left side
of the equation should be read “the probability of three successes,
(c)

given a process probability of 0.50 and three trials.”

In answering the probability questions, let us first list all the
possible outcomes of the three trials and compute the probabilities
The above probabilities can also be calculated using the equation.
You can try it yourself.
 Quantitative Techniques for Management Applications

134
Example 9.5 (Unsolved)

S
Notes
A very large lot of manufactured goods are to be sampled as a
___________________
check on its quality. Suppose it is assumed that 10 percent of the
___________________
items in the lot are defective and that a sample of 20 items is
___________________ drawn from the lot. What are the following probabilities:

E
___________________ 1. Probability of exactly zero defectives in the sample?
___________________ 2. Probability of more than one defective in the sample?
___________________
3. Probability of fewer than two defectives in the sample?

UP
___________________
Mean and Standard Deviation of the Binomial Distribution
___________________
The mean of a binomial distribution is found by multiplying the
___________________
probability of the event in which we are interested by n, the
___________________ number of trials:

Mean = np

This value is the same as the expected value as previously

discussed.
E-
The variance of a binomial distribution is calculated as:

Variance = npq so that the standard deviation is npε

Characteristics of the Binomial Distribution

1. It is a discrete distribution of the occurrences of an event with

CC

two outcomes - success or failure, good or bad.

2. The trials must be independent of one another. This
assumption implies sampling from an infinite population.
Sampling with replacement fulfils this requirement, but
where sampling without replacement is used, the binomial
distribution is still useful provided that the sample size is less
than 20.

3. As the number of trials grows and if p = 0.5 then the binomial

distributions approaches the normal distribution. For the
normal distribution to be an appropriate approximation, np
should be > 5.
(c)

Example 9.6 (unsolved)

Components are placed into bins containing 100. After inspection
of a large number of bins the average number of defective parts
was found to be 10 with a standard deviation of 3.
UNIT 9: Random Variables and Probability Distributions

135
Assuming that the same production conditions continue, except

S
Notes
that bins containing 300 were used: Activity
What is probability distribution
___________________
1. What would be the average number of defective components need for?
___________________
per larger bin?

E
___________________
2. What would be the standard deviation of the number of
___________________
defectives per larger bin?
___________________
3. How many components must each bin hold so that the
___________________
standard deviation of the number of defective components is

UP
equal to 1% of the total number of components in the bin? ___________________

___________________
Check Your Progress
___________________
Fill in the blanks:
___________________
1. The Excel and Quattro spreadsheet programs have a
function that can be used for evaluating binomial
probabilities, both __________ terms and __________
probabilities
E-
2. Mean = ______________

Poisson Distribution
The probability of getting specified number of successes from a
repeated number of trials can be obtained with the help of
CC

binomial distribution provided the probability of success or failure

is known. However, the probability of success or failure is finite in
this case. As the number of trials approaches infinity, Poisson
distribution is the limit of the binomial distribution.
Here c is the number of defects per sample, a is the expected
number of defects per sample, and e = 2.71828, the base of the
natural logarithm.
To explain this, we may think about the defects as being
distributed over the area of a surface, any unit being defective if it
contains one or more defects. If the surface area is divided into
very small units, so that no unit of the area has more than one
(c)

defect, the distinction between defect c and defective d disappears.

As the number n of units increases, nP remaining constant,
probability P must get smaller and smaller, and the binomial
distribution gradually approaches the Poisson.
 Quantitative Techniques for Management Applications

136 Table 9.2: Calculation of a Poisson Distribution for

S
Notes c = 0 through 5 and a = 0.5.

___________________

___________________

___________________

E
___________________

___________________

___________________

UP
___________________

___________________

___________________
The Poisson distribution is useful in ways other than as an
___________________
approximation to the binomial; for example, the analysis of queues,
or arrival and waiting line patterns at bridges and airports, tool
distribution points in factories, etc.

Characteristics of Poisson Distribution

E-
(a) It is a discrete distribution and is a limiting form of the
binomial distribution when n is large and p or q is small.
(b) Mean and variance are equal.
(c) It is usually, definitely positively skewed but cannot be
negatively skewed
CC

(d) As n becomes very large the Poisson distribution approximates

to the normal distribution
(e) The mean = np

Applications of Poisson Distribution

The Poisson distribution is similar to the binomial but
is used when n, the number of items or events, is large or unknown
and p, the probability of an occurrence, is very small relative to q,
the probability of non-occurrence. A rule of thumb is that the
Poisson distribution may be used when n is greater than 50 and
(c)

the mean np is less than 5. Some examples follow but it is

important to realize that the Poisson distribution only applies
when the events occur randomly, i.e., they are independent of one
another.
UNIT 9: Random Variables and Probability Distributions

Example 9.7 137

S
Notes
Customers arrive randomly at a service point at an average rate of
30 per hour. Assuming a Poisson distribution, calculate the ___________________
probability that: ___________________

(a) no customer arrives in any particular minute.

E
___________________

___________________
(b) exactly one customer arrives in any particular minute.
___________________
(c) two or more customers arrive in any particular minute.
___________________

UP
(d) Three or fewer customers arrive in any particular minute.
___________________
Solution:
___________________
The time interval to be used is one minute with a mean of 30/60 = ___________________
0.5
___________________
1. P (no customer) = 0.6065 from Table VI(a)

2. P (1 customer) = 0.3033 from Table VI(a)

3. P (2 or more) = 1 - 0.9098
E-
= 0.0902

The value of 0.9098 is the cumulative probability of 1 or

fewer customers arriving in a particular minute. As the
sum of the probability of every possible number of
arrivals equals 1, the probability of 2 or more = 1 - P(1 or
fewer).
CC

4. P(3 or fewer) = 0.9982

Example 9.8

A firm buys springs in very large quantities and from past records,
it is known that 0.2% is defective. The inspection department
sample the springs in batches of 500. It is required to set a
standard for the inspectors so that if more than the standard
number of defectives is found in a batch the consignment can be
rejected with at least 90% confidence that the supply is truly
defective.
(c)

How many defectives per batch should be set as the standard?

Solution:

With 0.2% defective and a sample size of 500 m= 500 × 0.2% = 1.

 Quantitative Techniques for Management Applications

138
To find the probability of 0, 1, 2, 3, etc. or more defectives the

S
Notes
respective probabilities are deducted from 1.
___________________

___________________
P(0 or more defectives) = certainty = 1

___________________ P(1 or more defectives) = 1 - 0.3679 = 0.6321

E
___________________ P(2 or more defectives) = 1 - 0.7358 = 0.2642
___________________ P(3 or more defectives) = 1 - 0.9197 = 0.0803
___________________
P(4 or more defectives) = 1 - 0.9810 = 0.0190

UP
___________________
These probabilities mean, for example, that there is a 26.42%
___________________
chance that 2 or more defectives will occur at random in a batch of
___________________ 500 with a 0.2% defect rate. If batches with 2 or more were rejected
then there can be 73.58% (1 - 0.2642) confidence that the supply is
___________________
defective.

As the firm wishes to be at least 90% confident, the standard

should be set at 3 or more defectives per batch. This level could
only occur at random in 8.03% of occasions so that
E-
the firm can be 91.97% confident that the supply is truly defective.

Check Your Progress

Fill in the blanks:
1. In Poisson distribution Mean and variance are
_________.
CC

2. The Poisson distribution is similar to the binomial but

is used when___________

The Normal Probability Distribution

Till recently we have been making curves which illustrated some of
the forms that a frequency distribution may assume. These curves
were based upon data of a few tens or hundreds of cases; each was
a sample drawn from a much larger, possibly infinite, universe.
Being a sample, a given curve would not necessarily have exactly
the same shape as the curve for the universe, but if the sample is
(c)

properly selected, the curve for the sample will tend to be of the
same general shape as the curve for the universe.
The normal curve represents a distribution of values that may
occur, under certain conditions, when chance is given full play. In
every case the necessary conditions include the existence of a large
UNIT 9: Random Variables and Probability Distributions

number of causes, each operating independently in a random 139

S
manner. Notes

Graphically, the normal distribution looks like a bell shaped curve: ___________________

There are certain properties of the normal distribution that you ___________________

can notice from the above curve:

E
___________________

1. It is symmetrical on both sides of the mean, i.e., mean, median ___________________

and mode coincide at the central value. ___________________

2. The curve never touches the x-axis and extends to -infinity on ___________________

UP
the left hand side to +infinity on the right hand side. ___________________
The normal distribution is an extremely important distribution. It ___________________
is easier to manipulate mathematically than many other
___________________
distributions and is a good approximation for several of the others.
In many cases, the normal distribution is a reasonable ___________________

approximation for binomial probability distribution for business

decision purposes; and in the following units, we shall use the
normal distribution in many of the applications. Despite its
general application, it should not be assumed that every process
E-
can be described as having a normal distribution.
The normal distribution is a function of z, the standard normal
variate, and is defined as:
Here the value of z is given by:
The normal distribution is completely determined by its expected
CC

value or mean (denoted by m) and standard deviation (s); that is,

once we know the mean and standard deviation, the shape and
location of the distribution is set. The curve reaches a maximum
at the mean of the distribution. One half of the area lies on either
side of the mean. The greater the value of s, the standard
deviation, the more spread out the curve?
With any normal distribution, approximately 0.50 of the area lies
within ±0.67 standard deviation from the mean; about 0.68 of the
area lies within ±1.0 standard deviations; and 0.95 of the area lies
with ±1.96 standard deviation.
(c)

Normal Probabilities on Spreadsheets

The Excel and Quattro spreadsheet function NORMSDIST
provides exactly the same values as in the Appendix 3. That is, for
any value of the standardized value of Z, it provides the left tail
cumulative normal probability. For example, for P(Z ≤ 0.67), use:
 Quantitative Techniques for Management Applications

140
NORMSDIST (0.67) = 0.7486

S
Notes

___________________
The next logical question might be: Given a normal distribution
with mean 4 and standard deviation s, what percentage of the total
___________________
area lies to the right or the left of a given X value? Alternatively,
___________________ we may ask: What is the probability of obtaining a value of X that

E
___________________ is as large as or larger than one specified? Since there are
infinitely many normal curves, each depending upon a particular
___________________
combination of mean and SD, the answer would vary from normal
___________________ curve to normal curve. The problem would be much less difficult if

UP
___________________ we could cause all the normal distributions to have the same mean
___________________
and SD.

___________________ This is exactly what we do when we standardize distributions; i.e.,

we cause them to have a mean of zero and a variance of one. Thus,
___________________
if we transform the given X value into a z value, where z is defined
as given above, and call Q(z) the probability of obtaining a value of
z that is equal to or larger than the one specified, tables are
already calculated to allow us to find Q(z), given z. Such a table is
given in Appendix 3.
E-
Example 9.9
Assume that your working hours X are distributed normally with
m = 5 and s = 2. What is the probability of your working 9 hours or
more than 9 hours?
CC

Solution:
First we standardize the specified X value: z=
Looking in the Appendix 3, we find that Q(z), the probability of
obtaining a value of z as large as or larger than specified, is
Q(z) = Q(2) = 0.02275
Alternatively, we may say that 2.275 per cent of the area in the
distribution is to the right of z = 2. This means that there is 2.275
per cent probability that you would be working for 9 or more than
9 hours in the day.
This could be looked at in another way. What is the probability
(c)

that you would be working less than 9 or 9 hours in a day? The

solution of this question can be found by subtracting the above
value from 1, so that we get the value which lies on the left of the 9
hour line and not on the right in the normal distribution. The
UNIT 9: Random Variables and Probability Distributions

answer is 0.97725 or 97.725 per cent probability that you would be 141

S
working for 9 or less than 9 hours on any given day. Notes

___________________
Example 9.10
An assembly line contains 2,000 components each one of which has ___________________

a limited life. Records show that the life of the components is

E
___________________
normally distributed with a mean of 900 hours and a standard
___________________
deviation of 80 hours.
___________________
1. What proportion of components will fail before 1,000 hours?
___________________
2. What proportion will fail before 750 hours?

UP
___________________
3. What proportion of components fails between 850 and 880
hours? ___________________

4. Given that the standard deviation will remain at 80 hours ___________________

what would the average life have to be to ensure that not
___________________
more than 10% of components fail before 900 hours?
Solution:
1. Here the value being investigated, 1000 hrs, is 1.25 standard
deviation away from the mean of 900 hours.
E-
If Appendix 3 is examined it will be seen that the value for a z
score of 1.25 is 0.3944. As one half of the distribution is less
than 900, the proportion which fails before 1,000 hours is 0.5 +
0.3944 = 89.44%.
If required this could be expressed as the number of
components which are expected to fail, thus
CC

2,000 × .8944 = 1788.8 which be rounded to 1789

2. From the tables in appendix 3 we obtain the value 0.4696. In
this case as we require the proportion that will fail before 750
hours, the table value is deducted from 0.5.
∴ Proportion expected to fail before 750 hours.
= 0.5 = 0.4696 = 0.0304, i.e., 3.04%
3. When it is required to find the proportion between two values
(neither of which is the mean) it is necessary to use the tables
to find the proportion between the mean and one value and
the proportion between the mean and the other value. Then
find the difference between the two proportions.
(c)

Which gives a proportion of 0.2340?

Which gives a proportion of 0.0987?
∴ Proportion between 850 and 880 is
0.2340 - 0.0987 = 0.1353, i.e., 13.53%
 Quantitative Techniques for Management Applications

142
Note: This part of the example illustrates the proportion between

S
Notes
two values on the same side of the means. If the two values are on
___________________
opposite sides of the mean, the calculated proportions would be
___________________ added.
___________________ This problem is the reversal of the earlier questions based on the

E
___________________ same principles. The earlier problems started with the mean and
standard deviation, found the z score and then the proportion from
___________________
the tables. We now start with the proportion and work back,
___________________ through the tables, to find a new mean value. If not more than

UP
___________________ 10% should be under 900 it follows that 90% of the area of the
___________________
curve must be greater than 900.

___________________ Bearing in mind that the tables only show values for half the
distribution (because both halves are identical) we have to look in
___________________
the tables for a value close to 0.4 (i.e. 0.9 - 0.5).
It will be seen that three is a value in the Table in Appendix 3 of
0.3997 i.e. virtually 0.4. This value has a z score of 1.28.
Thus
E-
∴ 102.4 = mean - 900
∴ mean = 1002.4 hours.
Thus if the mean life of the components is 1002.4 hours with a
standard deviation of 80 hours, less than 10% of the components
will fail before 900 hours.
CC

Hints on what Type of Distribution Applies

It is not always easy to recognize when a Normal, Binomial or
Poisson distribution should be used. The following hints might
help.

Binomial Distribution

(a) Outcomes have discrete values and do not have continuous

ranges of possible values. For example, in dealing with
people, there can only be discrete values; 1, 2, 3 etc., 2.35
people is not possible.
(c)

(b) There are only two possible conditions; good/bad, black/white,

male/female, acceptable/not acceptable and so on.
(c) The probability of an item having one of the two possible
conditions is p, thus the probability of the item having the
UNIT 9: Random Variables and Probability Distributions

other condition is (1 - p). (1 - p) is usually referred to as q, so 143

S
that (p + q) = 1. Notes
Activity
(d) When the number of items, n, is large and p is not close to 0 or What is the difference
___________________
1 so that the distribution is approximately symmetric, the between poisson distribution
and___________________
normal distribution?
binomial probabilities can be approximated using a Normal

E
___________________
Distribution with the same mean (m = np) and standard
deviation (s = Önpq). ___________________

___________________
Poisson distribution
___________________

UP
Similar to a binomial distribution, but used for rare events:
___________________
(a) The number of items, n, is large; say greater than 50.
___________________
(b) p is small in relation to q so that np (the mean of a Poisson ___________________
distribution) is less than, say, 5.
___________________
Normal Distribution

The most commonly applied probability distribution.

(a) It applies to variables with a continuous range of possible

E-
values. Examples include: time, weights, distances, sizes,
growth rates, etc.

(b) Where the quantities are large, the Normal Distribution can
also be used for discrete variables (see note above binomial
distribution).
CC

Check Your Progress

Fill in the blanks:
1. Poisson distribution is ______ to a binomial distribution,
but used for rare events.
2. Normal Distribution is applies to variables with a
________ range of possible values.

Summary
A probability distribution is a rule that assigns a probability to
(c)

every possible outcome of an experiment. An event whose

numerical value is determined by the outcome of an experiment is
called a variate or often a random variable. There are two kinds of
probability distributions, discrete and continuous. Discrete
probability distributions are those where only a finite number of
 Quantitative Techniques for Management Applications

144 outcomes are possible. Continuous probability distributions are

S
Notes those which represent continuously variable random variables.
___________________

___________________ Lesson End Activity

___________________ Two different digits are chosen at random from the set 1, 2, 3, 4, 5,

E
___________________ 6, 7, 8. Find the probability that sum of two digits exceeds 13.

___________________

___________________
Keywords

UP
___________________ Probability Distribution: A probability distribution is a rule that
assigns a probability to every possible outcome of an experiment.
___________________
Random Variable: An event whose numerical value is
___________________
determined by the outcome of an experiment is called a variate or
___________________
often a random variable.
Discrete Probability Distributions: Discrete probability
distributions are those where only a finite number of outcomes are
possible.
E-
Continuous Probability Distributions: Continuous probability
distributions are those which represent continuously variable
random variables

Questions for Discussion

1. Explain the meaning of conditional probability. State and
CC

prove the multiplication rule of probability of two events when

(a) they are not independent, (b) they are independent.
2. What is the probability of getting a sum of 2 or 8 or 12 in
single throw of two unbiased dice?
3. Two cards are drawn at random from a pack of 52 cards. What
is the probability that the first is a king and second is a queen?
4. What is the probability of successive drawing of an ace, a
king, a queen and a jack from a pack of 52 well shuffled cards?
The drawn cards are not replaced.
5. 5 unbiased coins with faces marked as 2 and 3 are tossed.
(c)

Find the probability of getting a sum of 12.

6. If 15 chocolates are distributed at random among
5 children, what is the probability that a particular child
receives 8 chocolates?
UNIT 9: Random Variables and Probability Distributions

145
7. A batch of 5,000 electric lamps has a mean life of 1,000 hours

S
Notes
and a standard deviation of 75 hours. Assume a Normal
Distribution. ___________________

___________________
(a) How many lamps will fail before 900 hours?

E
___________________
(b) How many lamps will fail between 950 and 1,000 hours?
___________________
(c) What proportion of lamps will fail before 925 hours?
___________________
(d) Given the same mean life, what would the standard ___________________

UP
deviation have to be to ensure that not more than 20% of
___________________
lamps fail before 916 hours?
___________________
8. A mail-order company is analyzing a random sample of its
computer records of customers. Among the results are the ___________________
following distributions: ___________________

Size of order Number of customers

No. of lubes April September
Less than 1 8 4
1 and less than 5 19 18
E-
5 and less than 10 38 39
10 and less than 15 40 69
15 and less than 20 22 41
20 and less than 30 13 20
30 and over 4 5
Total 144 196
CC

Required:

(a) Calculate the arithmetic mean and standard deviation

order size for the April sample;

(b) Find 95% confidence limits for the overall mean order size
for the April customers and explain their meaning;
(c) Compare the two distributions, given that the arithmetic
mean and standard deviation for the September sample
were 13.28 and 7.05 orders respectively.

9. The chief accountant of the Hotels Group is analyzing the

profitability of the group’s smallest hotel, the Unity. The
(c)

Unity has 120 bedrooms, each of which can be let as either a

single or a double room. The price of a single room is ` 45 per
night whereas for a double room it is ` 35 per person per
night. The hotel’s other main source of revenue is the
restaurant where the average price of an evening meal is ` 9.
 Quantitative Techniques for Management Applications

146
A proportion of residents can be expected to have an evening

S
Notes meal in the restaurant, and experience has shown that for the
___________________ Unity this proportion is usually about 60%.
___________________ Required:
___________________ (a) Develop an expression for the average daily revenue (R)

E
___________________ from residents at the hotel, including both the letting of
rooms and evening meals in the restaurant.
___________________

___________________
Assume that:

UP
___________________ i. the overall proportion of rooms occupied is p, and

___________________ ii. a proportion q of the rooms that are occupied are let as
single rooms.
___________________

___________________ (b) The proportion (q) of rooms which are let as single rooms is
usually between 10% and 30%, and it can be assumed that
this is independent of the overall proportion of rooms
occupied (p).

Determine the average daily revenue (R) in terms of the

E-
parameter p for values of q of 10%, 20% and 30% and plot
these three functions on a graph. What is the lowest
percentage occupancy (p) which will yield an average daily
revenue of ` 8,000, and for which of the given values of q
does this occur?

(c) What is the overall relationship between R and p if the

percentage of rooms let as single rooms is given by the
CC

following probability distribution?

q Probability

10% 0.3

20% 0.5

30% 0.2

(d) Alternatively, if q has a normal distribution with a mean

of 19% and a standard deviation of 7%, determine a 95%
confidence interval for R in terms of the parameter p.
(For the purposes of this calculation, you may ignore any
(c)

variability in the proportion of residents who take an

evening meal in the restaurant.)

10. Castrol Company Limited is planning to introduce a new

motor lube. The company’s marketing department estimates
that the prior distribution for likely sales is normal with a
UNIT 9: Random Variables and Probability Distributions

mean of 10,000 tonnes. In addition it has determined that 147

S
there is a probability of one half that the likely sales will lie Notes

between 8,000 and 12,000 tonnes. ___________________

The lube will sell for ` 10 per 100 gm but the publishing ___________________

company pays the parent company 10% of revenue in royalties

E
___________________
and the fixed costs of printing and marketing the lube are
___________________
calculated to be ` 25,000. Using current facilities, the variable
___________________
production costs are ` 4 per 100 gm. However, the Castrol
Company has the option of hiring a special machine for ` ___________________

UP
14,000 which will reduce the variable production costs to ` ___________________
2.50 per 100 gm.
___________________
Required: ___________________

(a) Show that the standard deviation of likely sales is ___________________

approximately s = 3,000.

(b) Using s = 3,000 determine the probability that the

company will at least break even if
E-
i) existing facilities are used,

ii) the special machine is hired.

(c) By comparing expected profits, decide whether or not the

company should hire the special machine.

(d) By using the normal distribution it can be shown that the

CC

following probability distribution may be applied to lube

sales:
Sales (‘000) 0-5 5-8 8-10 10-12 12-15 15-20
Probability 0.05 0.20 0.25 0.25 0.20 0.05

By assuming that the actual sales can only take the midpoints
of these classes, determine the expected value of perfect
information and interpret its value.
11. Assume that the probability of a salesperson making a sale at
a randomly selected petrol station is 0.1. If a salesperson
makes 20 calls a day, determine the following:
(c)

(a) The probability of no sales.

(b) The probability of one sale.

(c) The probability of four or more sales.

 Quantitative Techniques for Management Applications

148
(d) The probability of more than four sales.

S
Notes

___________________ (e) The probability of four sales.

___________________ 12. Daily sales of a certain product are known to have a normal
___________________ distribution of 20 per day, with a standard deviation of 6 per

E
day.
___________________

___________________ (a) What is the probability of selling fewer than 16 on a given

day?
___________________

UP
(b) What is the probability of selling between 15 and 25 units
___________________
on a given day?
___________________
(c) How many units would have to be in hand at the start of a
___________________
day in order to have less than a 10 percent chance of
___________________ running out?

13. On a midterm exam, the scores were distributed normally

with mean of 72 and standard deviation of 10. Student
Wright scored in the top 10 percent of the class on the
midterm.
E-
(a) Wright’s midterm score was at least how much?

(b) The final exam also had a normal distribution, but with
mean of 150 and standard deviation of 15. At least what
score should Wright get in order to keep the same ranking
(i.e., top 10 percent)?
CC

14. An investor wishes to invest in one of two projects. The

returns for both projects are uncertain, and the probability
distribution for returns can be expressed by a normal
distribution in each case. Project A has a mean return of `
240,000 with a standard deviation of ` 20,000. Project B has a
mean return of ` 250,000 and a standard deviation of ` 40,000.

(a) Consider a return of ` 280,000. Which project has a higher

chance of returning this much or more?

(b) Consider a return of ` 2,20,000. Which project has a

higher chance of returning this much or more?
(c)

15. A survey was conducted among the readers of a certain

magazine. The results showed that 60 percent of the readers
were homeowners and had incomes in excess of ` 25,000 per
month; 20 percent were homeowners but had incomes of less
than ` 25,000; 10 percent had incomes in excess of ` 25,000
UNIT 9: Random Variables and Probability Distributions

but were not homeowners; and the remaining 10 percent were 149

S
neither homeowners nor had incomes in excess of ` 25,000. Notes

(a) Suppose a reader of this magazine is selected at random ___________________

and you are told that the person is a homeowner. What is ___________________
the probability that the person has income in excess of

E
___________________
` 25,000?
___________________
(b) Are home ownership and income (measured only as above
___________________
or below ` 25,000) independent factors for this group?
___________________
16. A survey was conducted of families in an urban and the

UP
surrounding suburban area. The families were classified ___________________
according to whether or not they customarily watch two TV ___________________
programs. The data are shown in the table in percentages of
___________________
the total.
___________________
(a) If a family is selected from this group at random, what is
the probability that it views both programs?

(b) If the family selected views program A, what is the

probability that it also views program B?
E-
(c) Are the events (views program A) and (views program B)
independent events?

(d) Is the event (views program B) independent of the event

(urban)?

(e) Consider the event (view either program A or B or both). Is

CC

this event independent of the event (urban)?

Watch Watch Program A

Program B Yes No

Urban Suburban Urban Suburban Total

Yes 10% 14% 5% 1% 30%
No 15 21 20 14 70
Total 25% 35% 25% 15% 100%
17. Newspaper articles frequently cite the fact that in any one
year, a small percentage (say 10 percent) of all drivers is
responsible for all automobile accidents. The conclusion is
often reached that if only we would single out these accident-
(c)

prone drivers and either retrain them or remove them from

the roads, we could drastically reduce auto accidents. You are
told that of 100,000 drivers who were involved in one or more
accidents in one year, 11,000 of them were involved in one or
more accidents in the next year.
 Quantitative Techniques for Management Applications

150
(a) Given the above information, complete the entries in the

S
Notes
joint probability table below.
___________________

___________________
(b) Do you think searching for accident-prone drivers is an
effective way to reduce auto accidents? Why?
___________________

E
Table of accidents
___________________
Second Year
___________________
First Year Accident No Marginal Probability of
Accident Event in First Year
___________________

UP
Accident 0.10
___________________
No Accident 0.90
___________________ Marginal Probability of
Event in Second Year 0.10 0.90 1.00
___________________

___________________ 18. Gati India Ltd., maintains kilometre records on all of its
rolling equipment. Here are weekly kilometre records of its
trucks.

810 450 756 789 210 657 589 488 876 689
E-
1450 560 469 890 987 559 788 943 447 775

(a) Calculate the median kilometre a truck travelled.

(b) Calculate the mean for 20 trucks.

(c) Compare part (a) and (b) and explain which one is better
measure of central tendency of the data.
CC

19. Premier Automobiles Ltd. does statistical analysis for an

automobile racing team. Here are the fuel consumption
figures in kilometre per litre for the team’s cars in the recent
races.

4.77 6.11 6.11 5.05 5.99 4.91 5.27 6.01

5.75 4.89 6.05 5.22 6.02 5.24 6.11 5.02

(a) Calculate the median fuel consumption.

(b) Calculate the mean fuel consumption.

(c) Group the given data into equally sized classes. What is
(c)

the fuel consumption value of the modal classes.

(d) Which of the three measures of central tendency is best for

Premier to use when she orders fuel? Explain.

20. A machine is assumed to depreciate 40% in value in first year,

25% in second year and 10% per annum for the next three
UNIT 9: Random Variables and Probability Distributions

years. Each percentage is calculated on the diminishing 151

S
value. What is the average percentage depreciation for the Notes
five years? ___________________

21. Philips India Ltd., manufactures the famous Philips tube ___________________
lights of 40 watts. The company has developed a new variety

E
___________________
of Fluorescent 24 watt tube lights for specific applications in
control equipments used in defence components. Before it is ___________________
commercially launched the manager R and D desires to ensure ___________________
its reliability and quality. The test results conducted on 400
___________________
such tube lights are shown below. Compute the coefficient of

UP
skewness. ___________________

Life Time (hours) No. of Tubes ___________________

300-400 14 ___________________
400-500 46
___________________
500-600 58
600-700 76
700-800 68
800-900 62
E-
900-1000 48
1000-1100 22
1100-1200 6

22. From past experience it is known that a machine is set up

correctly on 90% of occasions. If the machine is set up
correctly then 95% of good parts are expected but if the
machine is not set up correctly then the probability of a good
CC

part is only 30%.

On a particular day the machine is set up and the first
component produced and found to be good. What is the
probability that the machine is set up correctly?
23. If the probability of obtaining heads when tossing a certain
coin is ½, what is the probability of obtaining heads four times
in nine tosses?
24. If the probability of obtaining a 6 when throwing a certain
dice is ½, what is the probability of obtaining a 6 four times in
nine throws?
25. In how many different ways the first three places can be filled
(c)

in a race in which there are 11 horses?

26. From our statisticians and seven engineers, a committee is to
be formed which must consist of two statisticians and two
engineers. In how many different ways, the committee can be
formed if (i) There is no other restriction on the membership
 Quantitative Techniques for Management Applications

152
of the committee. (ii) A particular statistician must be

S
Notes included, (iii) Two of the engineers are unable to serve on the
___________________ committee?

___________________
27. Five bolts are selected at random from a box containing six
sound and three faulty bolts. What is the probability of
___________________ obtaining (i) five sound, (ii) four sound and one faulty, (iii)

E
___________________ three sound and two faulty, (iv) two sound and three faulty
bolts?
___________________

___________________ Further Readings

UP
___________________ Books
___________________
R S Bhardwaj, Mathematics for Economics and Business, Excel
___________________ Books, New Delhi, 2005
___________________
D C Sanchethi and V K Kapoor, Business Mathematics

Sivayya and Sathya Rao, An Introduction to Business Mathematics

Web Readings
E-
www.managementstudyguide.com
www.textbooksonline.tn.nic.in
www.mathbusiness.com
CC
(c)
UNIT 10: Case Studies

Unit 10
153

S
Notes

Case Studies
___________________

___________________

E
___________________
Objectives
___________________
After analyzing these cases, the student will have an appreciation of the
concept of topics studied in this Block. ___________________

___________________

UP
___________________
Case Study 1: Selecting Distribution Type
___________________
In each of the following three situations, use the binomial,
Poisson or normal distribution according to which it is most ___________________
appropriate. In each case, explain why you selected the
___________________
distribution and draw attention to any feature which supports or
casts doubt on the choice of distribution.

Situation 1

The lifetimes of a certain type of electrical component are

E-
distributed with a mean of 800 hours and a standard deviation of
160 hours.

Questions

1. If the manufacturer replaces all components that fall before

the guaranteed minimum lifetime of 600 hours, what
percentage of the components has to be replaced?
CC

2. If the manufacturer wishes to replace only the 1% of

components that have the shortest life, what value should be
used as the guaranteed lifetime?

3. What is the probability that the mean lifetime of a sample of

25 of these electrical components exceeds 850 hours?

Situation 2

A green grocer buys peaches in large consignments directly from a

wholesaler. In view of the perishable nature of the commodity,
the green grocer accepts that 15% of the supplied peaches will
usually be unsaleable. As he cannot check all the peaches
individually, he selects a single batch of 10 peaches on which to
(c)

base his decision of whether to purchase a large consignment or

not. If no more than two of these peaches are unsatisfactory, the
greengrocer purchases the consignment.
Contd….
 Quantitative Techniques for Management Applications

154 Question

S
Notes
Determine the probability that, under normal supply conditions,
___________________ the consignment is purchased.
___________________ Situation 3
___________________ Vehicles pass a certain point on a busy single-carriageway road at

E
___________________
an average rate of two per ten-second interval.

___________________
Question

___________________
Determine the probability that more than three cars pass this
point during a twenty-second interval.

UP
___________________
Source: MBCQ-721D_Quantitative Techniques for Management Applications_CCE_UPES
___________________

___________________

___________________
E-
CC
(c)
UNIT 10: Case Studies

Case Study 2: Functional Probabilities 155

S
Thirty chief executive officers in an oil and gas industry are Notes
classified by age and by their previous functional position as ___________________
shown in the table below:
___________________
Previous Functional Age

E
___________________
position Under 55 55 and older Total
___________________
Finance 4 14 18

Marketing 1 5 6 ___________________
Other 4 2 6
___________________

UP
Total 9 21 30
___________________
Suppose an executive is selected at random from this group.
___________________
Questions
___________________
1. What is the probability that the executive chosen is under 55?
___________________
What type (marginal, conditional, joint) of probability is this?

2. What is the probability that an executive chosen at random is

55 or older and with Marketing as the previous functional
position? What type of probability is this?
E-
3. Suppose an executive is selected, and you are told that the
previous position was in Finance. What is the probability that
the executive is under 55? What kind of probability is this?

4. Are age and previous functional position independent factors

for this group of executives?
Source: MBCQ-721D_Quantitative Techniques for Management Applications_CCE_UPES
CC
(c)
 Quantitative Techniques for Management Applications

156

S
Notes

___________________

___________________

___________________

E
___________________

___________________

___________________

UP
___________________

___________________

___________________

___________________
E-
CC
(c)
UNIT 11: Decision Theory

157

S
Notes

___________________

___________________

E
___________________

___________________

___________________

___________________

UP
___________________

___________________

___________________

___________________

BLOCK-III
E-
CC
(c)
Detailed Contents Quantitative Techniques for Management Applications

158

S
Notes
UNIT 11: DECISION THEORY
___________________ UNIT 13: TRANSPORTATION MODELS
z Introduction z Introduction
___________________
z Key Terms z Demand, Transport Demand and Market Models
___________________

E
z Steps in Decision Theory z Types of Transportation Problem
___________________
UNIT 12: LINEAR PROGRAMMING UNIT 14: ASSIGNMENT PROBLEM
___________________
z Introduction z Introduction
___________________
Hungarian Method for Solving Assignment Problem

UP
z Linear Programming Problem z
___________________
z Graphical Method
___________________ UNIT 15: CASE STUDIES
z Simplex Method
___________________
z Duality
___________________
E-
CC
(c)
UNIT 11: Decision Theory

Unit 11
159

S
Activity
Notes
What is payoff means?

Decision Theory
___________________

___________________

E
___________________
Objectives
___________________
After completion of this unit, the students will be aware of the following
topics: ___________________

\ Key Terms ___________________

UP
\ Steps in Decision Theory ___________________
\ Decision Tree
___________________

___________________
Introduction
___________________
It assists the managers/executives in making the decisions. It deals
with the methods helpful to decision makers to select the best
course of action from amongst the alternative plans of action. It
provides a method of rational decision making when consequences
E-
are not fully known. It provides a framework for better
understanding of decision situations and for evaluating
alternatives.

Key Terms
Decision Maker: An individual/group responsible for making a
CC

choice of appropriate course of action.

Objectives: Which decision maker wants to achieve.
Preference/value System: Criteria that decision maker uses in
decision making a choice of best course of action.
Courses of Action: Decision alternatives under control available
to decision maker.
States of Nature: Decision maker prepares the list of possible
future events before applying theory. These future events are
called states of nature, are mutually exclusive and collectively
exhaustive. It can have/not have a numerical description (Low/high
(c)

demand of items, lockout or strike etc.)

Payoff: It is the effectiveness of particular combination of a course
of action and state of nature. These are also called as conditional
values/profits.
 Quantitative Techniques for Management Applications

160
Opportunity Loss: It is incurred due to failure of not adopting

S
Notes
Activity most favourable course of action or strategy.
What is scholastic situation?
___________________
Decisions: Broadly there are three types of decisions which are as
___________________ follows:
___________________
Strategic decisions are concerned with external environment of

E
___________________ the organization. (Selection of location, product, market,
technology, etc.)
___________________

___________________ Administrative Decisions are concerned with structuring and

UP
acquisition of the organization’s resources so as to optimize the
___________________
performance of the organization. (Layout, distribution channel,
___________________ purchase of assets, etc.)
___________________ Operating Decisions are primarily concerned with day to day
___________________ operations of the organization. (Production scheduling, inventory,
packing and dispatching).

Check Your Progress

Fill in the blanks
E-
1. Strategic decisions are concerned with ________
environment of the organization.
2. Administrative Decisions are concerned with
structuring and acquisition of the organization’s
___________
CC

Steps in Decision Theory

(i) Decision making environment

(ii) Objective of a decision maker

(iii) Alternative plans of actions or strategies

(iv) Decision Payoff

(i) Decision making environment

™ Deterministic situation: Where the information is
completely known and the outcome of a specified decision
(c)

can be predetermined with certainty. Decision maker has

the complete information of impact of each course of
action. The techniques used under such situations are: (i)
Linear Programming, (ii) Input-output analysis, (iii)
Transportation and assignment models (iv) CPM.
UNIT 11: Decision Theory

161
™ Stochastic situation: In this situation, the decision maker

S
Notes
knows the likelihood of the occurrence of each of state of
nature. The probability of occurrence of each state is ___________________
known. The techniques used are Decision Theory, PERT, ___________________
Simulation, Markov Chain, and Bayesian Theorem.

E
___________________
™ Situation of uncertainty: Wherein the probabilities
___________________
associated with the states of nature are unknown. For
___________________
example, the success of new product launched in the
market or the success of branch office opened abroad. ___________________

UP
Game Theory is used to analyse such situations. ___________________

(ii) Objective of a decision maker: The objective should be ___________________

defined explicitly. Whether he wants to continue the existing ___________________
state or switch over to other state. The problem relates to
___________________
single goal or multiple goals. Example: Maximizing return or
profit, minimizing loss or wastage.

(iii) Alternative plans of actions or strategies: There must be

more than one course of action; otherwise there is no need of
E-
any decision. Exhaustive list of alternatives is prepared and
then feasible alternatives should be considered for the
analysis.

(iv) Decision Payoff: It represents the effectiveness of the

strategies. Generally, it is measured in monetary terms. It can
be fixed in advance or can be random variable.
CC

Payoff Table: List of states of nature (events) which are mutually

exclusive and collectively exhaustive and a set of given courses of
action (strategies) and for each combination of these, payoff is
calculated.
Table 11.1: Payoff Table

State of nature Courses of action

S1 S2 S3 ……… Sn
O1 a11 a12 a13 a1n
O2 a21 a22 a23 …… a2n
O3 a31 a32 a33 ……. a3n
. .. .. .. ..
(c)

Om am1 am2 am3 ……. amn

EMV (Expected Monetary Value): For a given course of action is

the weighted average payoff, which is the sum of the product of
payoff for the several combinations of strategies and states of
 Quantitative Techniques for Management Applications

162
nature, multiplied by the probability of occurrence of each

S
Notes outcome.
___________________
Steps for Calculating EMV
___________________
(i) Define systematically the states of nature (Oi) and course of
___________________

E
action (Sj).
___________________
(ii) List the payoff associated with each combination of state of
___________________
nature and course of action along with probability of each
___________________ state of nature.

UP
___________________ (iii) Calculate EMV for each course of action by multiplying the
___________________ conditional payoffs by associated probabilities and sum up
these weighted values for each course of action.
___________________

___________________ (iv) On the basis of specified decision objective, determine the

courses of action corresponding to optimal EMV.

EPPI (Expected Profit with Perfect Information): It is the

maximum obtainable expected monetary value based on perfect
information as to which state of nature will occur.
E-
EVPI (Expected Value of Perfect Information): It is the
maximum amount one would be willing to pay to obtain perfect
information.

EVPI = EPPI – EMV

EOL (Expected Opportunities Loss): It occurs due to lack of

CC

perfect information. It shows the expected differences between the

payoff of right decision (maximum) and payoff of actual decision.

EOL = Summation of COL (Oi, Sj). P(Oi)

Where COL is the conditional opportunity loss

Decision Tree
This approach is used for complicated situations. A decision tree is
a decision flow diagram that includes branches leading to
alternatives one can select among the usual branches leading to
events that depend on probabilities. Expectation principle is used
(c)

i.e. to choose the alternative that maximizes the expected profit or

minimizes the expected cost.

Advantages

(i) It provides a graphical presentation of sequential actions or

decisions.
UNIT 11: Decision Theory

(ii) It makes the analysis simple as the computed values can be 163

S
written on the tree diagram. Notes
Activity
What is decision tree meant
(iii) It clearly depicts when decisions are expected to be made for?
___________________

along with their possible consequences and results. ___________________

(iv) The complex problems can be solved with tree diagram.

E
___________________

___________________
Limitations
___________________
(i) The probability estimates may not be accurate thus, may not
___________________
give true results.

UP
___________________
(ii) Certain extraneous variables may be out of control of the
___________________
decision maker and thus, objective may not be achieved
completely. ___________________

(iii) When number of states of nature is large, the problem ___________________

becomes complicated.

Check Your Progress

Fill in the blanks
E-
1. The probability estimates ______ be accurate
2. A decision tree is a _______________ that includes
branches leading to alternatives one can select among
the usual branches leading to events that depend on
probabilities
CC

Summary
Decision theory provides a framework for better understanding of
decision situations and for evaluating alternatives. Broadly there
are three types of decisions including Strategic decisions,
Administrative Decisions and Operating Decisions.
A decision tree is a decision flow diagram that includes branches
leading to alternatives one can select among the usual branches
leading to events that depend on probabilities.
(c)

Lesson End Activity

The marketing department of a certain petrochemicals
manufacturing company has worked out the payoff (given in the
following table) in terms of profit expressed in million dollars,
 Quantitative Techniques for Management Applications

164 concerning a technical proposal depending upon the rate of

S
Notes technological changes.
___________________
Technological Course of Action
___________________ Changes Accept Reject
___________________ Fast 2 3

E
___________________ Slow 5 2

___________________ None –1 4

___________________ Whether the technical proposal should be accepted or not?

UP
___________________
Keywords
___________________
Decision Maker: An individual/group responsible for making a
___________________
choice of appropriate course of action.
___________________
Payoff: It is the effectiveness of particular combination of a course
of action and state of nature. These are also called as conditional
values/profits.

Decision Tree: A decision tree is a decision flow diagram that

E-
includes branches leading to alternatives one can select among the
usual branches leading to events that depend on probabilities.

EPPI (Expected Profit with Perfect Information): It is the

maximum obtainable expected monetary value based on perfect
information as to which state of nature will occur.

EVPI (Expected Value of Perfect Information): It is the

CC

maximum amount one would be willing to pay to obtain perfect

information.

Questions for Discussion

1. What are the key characteristics of decision theory?
2. What are the different types of strategic decisions?
3. What are the key steps in decision theory?
4. Illustrate the calculation of EMV.
5. Define the approach of decision tree. What are the key
(c)

advantages and disadvantages of decision tree approach?

6. A distributor Doon Enterprises buys tankers of diesel for

` 10,00,000 and sells them at ` 12,00,000 each. All the tankers
left at the end of the day are worthless. Over the last 100 days
UNIT 11: Decision Theory

he recorded the daily sale and the distribution of sale is as 165

S
follows: Notes

Tankers Sold No. of Days ___________________

20 5
___________________
21 20

E
22 30 ___________________

23 35 ___________________
24 10
___________________
Determine the number of tankers the distributor should buy
___________________
so as to maximize his profit.

UP
___________________
7 An auto parts retailer sells headlights for ` 35 each which he
buys at ` 30 each. He cannot return the unsold headlights. ___________________
The daily demand has the following distribution: ___________________
No. of Headlights 23 24 25 26 27 28 29 30 31 ___________________
Probability 0.01 0.03 0.06 0.1 0.2 0.25 0.1 0.050 0.05

If each day’s demand is independent of previous day’s

demand, how many headlights should he order each day?
E-
8. An Oil exploration company wants to make a decision
regarding buying the site on lease or not, the cost has been
calculated for these two states for getting oil or not getting oil.
Draw a decision tree for the data given in the cost matrix:
State of Nature Getting Oil Not getting Oil
Prob. 0.01 0.99
Go for Lease ` 1,00,000 ` 1,00,000
CC

Not go for Lease ` 80,00,000 `0

Also advise the company which decision it should take.

(Ans. ` 80,000)

9. A company owns a lease on a certain property in a European

country. It may sell the lease for US $ 18,000 or it may drill
this property for exploring gas. Various possible drilling
results were obtained after a research conducted by its
engineers. The data is given in the following table:
Possible Results Probabili Outcomes
ty (US$)
(c)

Dry well 0.10 - 1,00,000

Oil well 0.20 45,000
Gas well only 0.40 98,000
Oil and gas combination 0.30 1,99,000
 Quantitative Techniques for Management Applications

166
Construct a decision tree diagram for the above problem and

S
Notes
determine EMV for the company. What will you suggest the
___________________
company to sell or drill?
___________________
10. A medical practitioner purchases a specific vaccine on Sunday
___________________
evening each week. The vaccine has to be used in the following

E
___________________ week otherwise it becomes worthless. It costs ` 20 per dose
and he charges ` 22 only per dose. The practitioner
___________________
administered the vaccine in following quantities in last 50
___________________ weeks:

UP
___________________ Number of weeks 20 25 40 60

___________________ Number of doses/ week 5 15 25 5

___________________ Calculate the number of doses the practitioner must buy every
___________________
week.

Further Readings

Books
E-
R S Bhardwaj, Mathematics for Economics and Business, Excel
Books, New Delhi, 2005

D C Sanchethi and V K Kapoor, Business Mathematics

Sivayya and Sathya Rao, An Introduction to Business Mathematics

Web Readings
CC

www.textbooksonline.tn.nic.in
www.mathbusiness.com
(c)
UNIT 12: Linear Programming

Unit 12
167

S
Activity
Notes
What is L.P?

Linear Programming
___________________

___________________

E
___________________
Objectives
___________________
After completion of this unit, the students will be aware of the following
topics: ___________________

\ The significance and advantages of Linear Programming ___________________

UP
\ The conversion of the LP problem into mathematical model ___________________
\ How to solve the LP problems of minimization and maximization using
graphical and simplex methods ___________________

\ How to obtain the dual of primal ___________________

___________________

Introduction
It was introduced by George Dantzig in 1947. It is used as a
scientific approach to decision making and an optimization
E-
technique used in operations research. It is applied to minimize
the cost or maximize the profit.

Linear Programming problem

A Linear Programming problem is a special case of a
Mathematical Programming problem. From an analytical
CC

perspective, a mathematical program tries to identify an extreme

(i.e., minimum or maximum) point of a function, which
furthermore satisfies a set of constraints, e.g., Linear programming
is the specialization of mathematical programming to the case
where both, function Z - to be called the objective function - and the
problem constraints are linear.
It allows the rationalization of many managerial and/or
technological decisions required by contemporary techno-socio-
economic applications. An important factor for the applicability of
the mathematical programming methodology in various
(c)

application contexts is the computational tractability of the

resulting analytical models. Under the advent of modern
computing technology, this tractability requirement translates to
the existence of effective and efficient algorithmic procedures able
to provide a systematic and fast solution to these models.
 Quantitative Techniques for Management Applications

168
Let us discuss the meaning of terms Linear and Program used in

S
Notes LP:
___________________
Linear: The relationship between the variables is directly
___________________ proportional. For example, if a wooden table requires 30 cubic feet
___________________ of wood; then 10 tables would require 300 cubic feet wood.

E
___________________ Program: A program is a set of instructions arranged in a logical
sequence.
___________________

___________________ Optimal: It means if a programme maximizes or minimizes some

UP
measure or criterion of effectiveness. Ex. Maximization of profit /
___________________
sales or minimization of cost or distance etc.
___________________
Limited: Availability of resources during planning horizon.
___________________
The related problem of Integer Programming requires some or all
___________________
of the variables to take integer (whole number) values. Integer
programs (IPs) often have the advantage of being more realistic
than LPs, but the disadvantage of being much harder to solve. The
most widely used general-purpose techniques for solving IPs use
the solutions to a series of LPs to manage the search for integer
E-
solutions and to prove optimality. Thus, most IP software is built
upon LP software.

Linear programming has proved valuable for modelling many and

diverse types of problems in planning, routing, scheduling,
assignment, and design. Industries that make use of LP and its
extensions include transportation, energy, telecommunications,
CC

and manufacturing of many kinds.

Check Your Progress

Fill in the blanks
1. Linear programming is the specialization of
mathematical programming to the case where both
function Z - to be called the ___________function and the
problem constraints are linear.
2. ____________means maximisation of profit or
minimisation of cost of a business subject to constraints
(c)

or hurdles.

Assumptions
Certainty- in Linear Programming, it is assumed that all model
parameters such as available resources, coefficients of objective
UNIT 12: Linear Programming

function and coefficients of the constraints are known with 169

S
certainty. Notes

Proportionality- all relationships in objective functions and ___________________

constraints are linear. In economic terminology, this means that ___________________

there are constant returns to scale i.e., if one unit of a product

E
___________________
contributes `10 as profit, then five units will contribute ` 50.
___________________
Additivity- the total of all the activities is given by sum total of
___________________
each activity performed separately. For example, the total profit in
the objective function will be sum of the profits contributed by each ___________________

UP
of products separately. ___________________

Continuity- means that the decision variables are continuous. ___________________

Accordingly, the solutions of decision variables and resources are
___________________
assumed to have whole numbers or fractions. When only integers
are required such as the number of tables/books/workers/machines, ___________________
Integer Programming is more suitable.

Finite Choices- implies that finite numbers of choices are

available to a decision maker and decision variables do not assume
E-
negative values.

Advantages
It is used to determine the product mix.
It helps in attaining the optimum use of production factors.
It gives possible and practical solutions, optimal solution for the
CC

decision maker.
It improves the quality of decisions as these are objective in
nature.
It is also helpful in re-evaluation of a basic plan for changing
conditions.

Limitations
Linearity: It treats all relationships among decision variables and
objective functions as linear but in real world situation it may not
be so.
(c)

Integer Value: The solution of LPP may not result into integer
values, the values are in fractions, rounding off to the nearest
integers does not give the optimal solution. Integer Programming
is the modified LPP which overcomes this limitation.
 Quantitative Techniques for Management Applications

170
Multiple Goals: It deals with only single objective (Minimizing

S
Notes
the cost or maximizing the profit) but practically a manager has to
___________________
make decision considering various goals at the same time i.e. sales
___________________ maximization, minimizing idle labour time and maximizing
___________________ capacity utilization etc. Goal Programming is the technique which

E
is suitable for such situations.
___________________
Constant Parameters: All the parameters are assumed to be
___________________
constant in LPP but it is not so in real life situations.
___________________
Number of Variables: It can be solved manually when there are

UP
___________________
two or three variables, in case of large number of variables
___________________ computing can be done through software only.
___________________ Deterministic Nature: It is assumed that all the coefficients and
___________________ resources are known with certainty and it is not affected by time
and uncertainty.
Problem Formulation (Mathematical Model): When a decision
has to be taken regarding the number of units of two or more than
two products to be manufactured under certain constraints like
E-
manpower, material or space etc., the mathematical or analytical
model of linear programming has to be used. There are three
components of mathematical model:

(i) Objective Function (Z): The linear function which is to be

optimized is called Objective Function. The objective functions
can be in terms of minimization or maximization. In case of
CC

minimization, we can have cost, time, distance etc. whereas in

case of maximization, these can be profit, revenue or sales etc.
For example:
Max. Z = 100 X1 + 200X2

Where X1 and X2 are the quantities (number of units) of two

products P1 and P2

`100 and `200 are the profit values for these two products P1
and P2.

(ii) Constraint (Inequality) Equations: These represent the

(c)

linear relationship of constraints in a given situation. These

can be of types such as: material, manpower, space, machine
or budget etc. The equations will have the signs less than
equal to (<=), greater than equal to (>=) or equality (=).
2X1 + 5X2 <= 2000
UNIT 12: Linear Programming

171
2 and 5 are the quantities of material consumed (cubic feet for

S
Notes
wooden material) for P1 and P2.
___________________
2000 is the total quantity of material available.
___________________
3X1 + 4X2 <= 4200

E
___________________

3 and 5 are the man-hours for P1 and P2. ___________________

4,000 is the total man-hours available. ___________________

___________________
(iii) Non-negative Equations: The decision variables have the

UP
values zero or positive, not negative. Thus, these are also ___________________
called as variable sign restrictions. ___________________
X1 and X2 >= 0 ___________________

Check Your Progress ___________________

Fill in the blanks:

1. All the parameters are assumed to be ________ in LPP
but it is not so in real life situations.
E-
2. The solution of LPP may not result into __________
values.

Graphical Method
Linear programming problems with two variables can be
CC

represented and solved graphically with ease. Though in real-life,

the two variable problems are practiced very little, the
interpretation of this method will help to understand the simplex
method. The solution method of solving the problem through
graphical method is discussed with an example given below.
Example 12.1: A company manufactures two types of boxes,
corrugated and ordinary cartons. The boxes undergo two major
processes: cutting and pinning operations. The profits per unit are
` 6 and ` 4 respectively. Each corrugated box requires 2 minutes
for cutting and 3 minutes for pinning operation, whereas each
carton box requires 2 minutes for cutting and 1 minute for pinning.
(c)

The available operating time is 120 minutes and 60 minutes for

cutting and pinning machines. Determine the optimum quantities
of the two boxes to maximize the profits.
 Quantitative Techniques for Management Applications

172
Solution:

S
Notes
Key Decision
___________________
To determine how many (number of) corrugated and carton boxes
___________________
are to be manufactured.
___________________

E
Decision Variables
___________________
Let xl be the number of corrugated boxes to be manufactured.
___________________
Let x2 be the number of carton boxes to be manufactured
___________________

UP
Objective Function
___________________
The objective is to maximize the profits. Given profits on
___________________
corrugated box and carton box are ` 6 and ` 4 respectively.
___________________
The objective function is,
___________________
Zmax = 6x1 + 4x2

Constraints
The available machine-hours for each machine and the time
consumed by each product are given.
E-
Therefore, the constraints are,
2x1 + 3x2 ≤ 120 (i)

2x1+ x2 ≤ 60 (ii)

where x1, x2 ≥ 0
CC

Graphical Solution
As a first step, the inequality constraints are removed by replacing
equal to sign to give the following equations:
2x1 + 3x2 = 120 (1)

2x1 + x2 = 60 (2)

Find the coordinates of the lines by substituting x1 = 0 and x2 = 0 in

each equation. In equation 1, put x1 = 0 to get x2 and vice versa

2x1 + 3x2 = 120

2(0) + 3x2 = 120, x2 = 40

(c)

Similarly put x2 = 0,

2x1 + 3x2 = 120

2x1 + 3(0) = 120, x1 = 60

UNIT 12: Linear Programming

173
The line 2x1 + 3x2 = 120 passes through coordinates (0, 40) (60, 0).

S
Notes
The line 2x1 + x2 = 60 passes through coordinates (0, 60) (30, 0). ___________________

The lines are drawn on a graph with horizontal and vertical axis ___________________
representing boxes x1 and x2 respectively. Figure 12.1 shows the

E
___________________
first line plotted.
___________________

___________________

___________________

UP
___________________

___________________

___________________

___________________
E-
Figure 12.1: Graph Considering First Constraint

The inequality constraint of the first line is (less than or equal to) ≤
type which means the feasible solution zone lies towards the
origin. The no shaded portion can be seen is the feasible area
shown in Figure 12.2 (Note: If the constraint type is ≥ then the
solution zone area lies away from the origin in the opposite
CC

direction). Now the second constraints line is drawn.

(c)

Figure 12.2: Graph Showing Feasible Area

 Quantitative Techniques for Management Applications

174
When the second constraint is drawn, you may notice that a

S
Notes
portion of feasible area is cut. This indicates that while considering
___________________ both the constraints, the feasible region gets reduced further. Now
___________________ any point in the shaded portion will satisfy the constraint
equations. For example, let the solution point be (15, 20) which lies
___________________

E
in the feasible region.
___________________
If the points are substituted in all the equations, it should satisfy
___________________ the conditions.
___________________ 2x1 + 3x2 ≤ 120 = 30 + 60 ≤ 120 = 90 ≤ 120

UP
___________________ 2x1 + x2 ≤ 60 = 30 + 20 ≤ 60 = 50 ≤ 60
___________________
Now, the objective is to maximize the profit. The point that lies at
___________________ the furthermost point of the feasible area will give the maximum
___________________ profit. To locate the point, we need to plot the objective function
(profit) line.
Equate the objective function for any specific profit value Z,
Consider a Z-value of 60, i.e.
E-
6x1 + 4x2 = 60
Substituting x1 = 0, we get x2 = 15 and
if x2 = 0, then x1 = 10

Therefore, the co-ordinates for the objective function line are (0, 15),
(10, 0) as indicated by dotted line L1 in figure. The objective function
line contains all possible combinations of values of xl and x2.
CC

The line L1 does not give the maximum profit because the
furthermost point of the feasible area lies above the line L1. Move
the line (parallel to line L1) away from the origin to locate the
furthermost point. The point P is the furthermost point, since no
area is seen further. Take the corresponding values of x1 and x2
from point P, which is 15 and 30 respectively, and are the optimum
feasible values of x1 and x2.

Therefore, we conclude that to maximize profit, 15 numbers of

corrugated boxes and 30 numbers of carton boxes should be
produced to get a maximum profit. Substituting x1 = 15 and x2 = 30
(c)

in objective function, we get

Zmax = 6x1 + 4x2

= 6(15) + 4(30)
Maximum profit: ` 210.00
UNIT 12: Linear Programming

175
Check Your Progress

S
Notes
Activity
Fill in the blanks: When do we use simplex
___________________
method?
1. Linear programming problems with two variables can be
___________________
represented and solved ___________ with ease.

E
___________________
2. Linear programming methods can be applied to various
___________________
fields such as production, marketing, sales, banking and
farm management, to solve the problems arising due to ___________________
limited resources and ___________ of resources. ___________________

UP
___________________
Simplex Method
___________________
Two families of solution techniques are in wide use today. Both ___________________
visit a progressively improving series of trial solutions, until a
___________________
solution is reached that satisfies the conditions for an optimum.
Simplex methods visit “basic” solutions computed by fixing enough
of the variables at their bounds to reduce the constraints to a
square system, which can be solved for unique values of the
E-
remaining variables. Basic solutions represent extreme boundary
points of the feasible region. The simplex method can be viewed as
moving from one such point to another along the edges of the
boundary.

Barrier or interior-point methods, by contrast, visit points within

the interior of the feasible region. These methods derive from
CC

techniques for nonlinear programming that were developed and

popularized in the 1960s by Fiacco and McCormick, but their
application to linear programming dates back only to Karmarkar’s
innovative analysis in 1984.

Box 12.1: Simplex Method for Standard Maximization Problem

Step 1: Convert to a system of equations by introducing slack
variables to turn the constraints into equations, and rewriting the
objective function in standard form.
Step 2: Write down the initial tableau.
Step 3: Calculate Zj and then Cj and then calculate Cj–Zj. Select
the pivot column: Choose the positive number with the largest
(c)

magnitude in the index row. Its column is the pivot column. (If
there are two candidates, choose either one.) If all the numbers in
the bottom row are zero or negative (excluding the rightmost
entry- Minimum Ratio), then the solution obtained is the feasible
solution.
Contd….
 Quantitative Techniques for Management Applications

176
Step 4: Select the pivot in the pivot column: The pivot must

S
Notes
Activity
always be a positive number. For each positive entry b in the
What is duality?
___________________ pivot column, compute the ratio a/b, where a is the number in the
___________________ Answer column in that row. Of these test ratios, choose the
smallest one. The corresponding number b is the pivot.
___________________

E
Step 5: Divide the pivot row by pivot element to make it unity.
___________________
Construct the new tableau by writing the previous pivot row first
___________________ at the same position (as it was having previously).

___________________ Step 6: Write the values of other rows so as the corresponding

UP
element in the pivot column becomes zero. Again calculate Zj and
___________________
Cj–Zj and iterate step 3 onwards to reach the feasible solution.
___________________
Check Your Progress
___________________
Fill in the blanks
___________________
1. Two families of _________________ techniques are in
wide use today.
2. Interior-point methods, by contrast, visit points within
the interior of the ________________ region.
E-
3. The concept of ________________ was introduced by
W.W. Leontief, who was of the opinion that the variables
are in a linear relationship with each other and
therefore, a problem can be stated in mathematical
terms.
CC

Duality
For every given linear programming problem, there exists an
intimately related L.P. Problem referred to as its Dual. The given
(original) problem is known as Primal. The duality theorem states
that “for every maximization (minimization) problem there is a
unique similar problem of minimization (maximization) involving
the same data which describes the original problem”. The ‘DUAL’
of a ‘DUAL’ is PRIMAL.

The characteristics of dual problem:

(c)

z If the objective of the primal is maximization, the objective of

the dual is minimization.
z The primal has m-constraints while its dual has m-unknowns
and vice-versa.
UNIT 12: Linear Programming

177
z The coefficients of the objective function of primal become the

S
Notes
constraints of its dual and vice-versa.
___________________
z The variables of the primal are replaced by the new variables
of its dual. ___________________

The sign of the inequalities in the set of restrictions of the

E
z ___________________
primal (<=) is reversed in the set of restrictions in its dual ___________________
(>=).
___________________
z For finding the dual of the given maximization problem, all the
___________________
constraint inequalities should be of (<=) type and for

UP
___________________
minimization, these should be of (>=) type.
___________________
Example 12.2: For obtaining the dual of following primal problem:
Max. Z = 3X1 + X2 + 2X3 – X4 ___________________

St: 2X1 – X2 + 3X3 + X4 = 10 ___________________

X1 + X2 – X3 + X4 = 11
X1, X2 >= 0, X3, X4 unrestricted in sign
For coefficients of objective function, the matrix
E-
is: [3 1 2 -1]
Another matrix for coefficients and resources are:
[2 -1 3 1] [X1] <= [10]
[1 1 -1 1] [X2] [11]
The variables X3 and X4 in Primal are restricted in sign; therefore,
CC

the third and fourth constraints in the dual will have equality sign.
As both the constraints in primal are of equality sign,
corresponding dual variables will be unrestricted in sign.
Let W1, W2 be the corresponding dual variables. The dual is as
follows:
[2 1]
[-1 1] [W1 W2] <= [10]

[3 –1] [11]
[1 1]
2W1 + W2 >= 3
(c)

–W1 + W2 >= 1
3W1 – W2 = 2
W1 + W2 = -1
W1, W2 are unrestricted in sign.
 Quantitative Techniques for Management Applications

178
Check Your Progress

S
Notes
Fill in the blanks:
___________________

___________________
1. For every given __________ programming problem, there
exists an intimately related L.P.
___________________

E
2. The coefficients of the __________ function of primal
___________________
become the constraints of its dual and vice-versa.
___________________
3. __________ introduced the concept of linear
___________________
programming, which is one of the most widely used

UP
___________________ methods of operations research.
___________________ 4. In linear programming, the problems can be solved by
___________________ using linear programming methods such as
___________and graphical, which aim at either
___________________
maximising the profit or minimising the cost.

Summary
A Linear Programming problem is a special case of a Mathematical
E-
Programming problem. It allows the rationalization of many
managerial and/or technological decisions required by contemporary
techno-socio-economic applications. Linear programming has proved
valuable for modelling many and diverse types of problems in
planning, routing, scheduling, assignment, and design.
Two families of solution techniques are in wide use today. Both
CC

visits a progressively improving series of trial solutions, until a

solution is reached that satisfies the conditions for an optimum.
Simplex methods visit "basic" solutions computed by fixing enough
of the variables at their bounds to reduce the constraints to a
square system, which can be solved for unique values of the
remaining variables. Barrier or interior-point methods, by
contrast, visit points within the interior of the feasible region.

Lesson End Activity

Which two families of solution techniques are in wide use today?

(c)

Keywords
Linear: The relationship between the variables is directly
proportional. For example, if a wooden table requires 30 cubic feet
of wood then 10 tables would require 300 cubic feet wood.
UNIT 12: Linear Programming

Program: A program is a set of instructions arranged in a logical 179

S
sequence. Notes

Optimal: It means if a programme maximizes or minimizes some ___________________

measure or criterion of effectiveness. Ex. Maximization of profit / ___________________

sales or minimization of cost or distance etc.

E
___________________
Objective Function (Z): The linear function which is to be ___________________
optimized is called Objective Function.
___________________

Questions for Discussion ___________________

UP
___________________
1. Define a linear programming problem. What are the key
assumptions of a linear programming problem? ___________________

___________________
2. What are the key advantages and limitations of LP?
___________________
3. What are the key components of mathematical model?
4. State the difference between the simplex methods and barrier
or interior point methods.
5. Write a note on duality theorem.
E-
6. A home decorator Glamour Enterprises manufactures two
types of lamps which go under two first technicians, a cutter,
second a finisher. Lamp A requires 2 hours of cutter’s time
and 1 hour of finisher’s time, Lamp B requires 1 hour of
cutter’s time and 2 hour of finisher’s time. The cutter has 104
hours and finisher 76 hours of time available each month.
CC

Profit on one lamp A is ` 6 and on one lamp B is ` 11.

Assuming that he can sell all the lamps he produces,
formulate the problem.

7. A firm manufactures three products A, B and C. The profits

are ` 30, ` 20 and ` 40 respectively. The firm has two
machines M1 and M2 and required processing time in minutes
on each product in given below:
Product
Machine A B C
M1 4 3 5
M2 2 2 4
(c)

Machines M1 and M2 have 2,000 and 2,500 machine minutes

respectively. The firm manufactures 100 A’s, 200 B’s and 50
C’s but not more than 150 A’s. Formulate the problem as a
LPP.
 Quantitative Techniques for Management Applications

180
8. The objective of a diet problem is to ascertain the quantities of

S
Notes foods that should be eaten to meet certain nutritional
___________________ requirement at a minimum cost. The consideration is limited
to milk, beef and eggs, and to vitamins A, B and C. The
___________________
number of grams of each of these vitamins contained in a unit
___________________ of each food is given below:

E
___________________ Vitamin Gallon of Pound of Dozens of Minimum daily
Milk Beef Eggs requirement
___________________
A 1 1 10 1
___________________ B 100 10 10 50

UP
___________________ C 10 100 10 10
Cost 40 90 20
___________________
(Rs.)
___________________
Formulate the mathematical model.
___________________
9 A textile unit has two grades of inspectors, I and II, who are to
be assigned for the quality control inspection. It is required
that 2,000 pieces be inspected per 8 hours a day. Grade I
inspectors can check pieces at the rate of 50 per hour with an
E-
accuracy of 97%, and grade II inspectors can check pieces at
the rate of 40 per hour with an accuracy of 95%. The wage
rate of grade I inspectors is ` 4.50 per hour and that of grade
II is ` 2.50 per hour. Each time an error is made by an
inspector, the cost to the company is one rupee. The company
has available I 10 grade 1 and 5 grade II inspectors for the
inspection job. Formulate the problem to minimize the total
CC

cost of inspection.
10. (a) Solve the following problems graphically:
Max Z = 20 X1 + 30 X2

Sub. to:
X1 + X2 <= 1
3X1 + X2 <= 4
X1, X2 >= 0 (Ans. X1=0, X2=1, Z=30)
(b) Max Z = 30 X1 + 50 X2

Subject to:
(c)

X1 + 2X2 <= 2000

X1 + X2 <= 1500
X2 <= 600
X1, X2 >= 0 (Ans. X1=1000, X2=500, Z=55,000)
UNIT 12: Linear Programming

181
(c) Max. Z = 4X1 + 5X2

S
Notes
Subject to:
___________________
X1 + X2 >= 1
___________________
-2X1 + X2 <= 1

E
___________________
4X1–2X2 <= 1 ___________________
X1, X2 >= 0 (Ans. Unbounded Solution) ___________________
(d) Min. Z = -X1 + 2X2 ___________________

UP
Subject to: ___________________

-X1 + 3X2 <= 10 ___________________

X1 + X2 <= 6 ___________________

X1 – X2 <= 2 (Ans. X1=2, X2=0, Z= - 2) ___________________

(e) Min. Z = 20 X1 + 10 X2

Subject to:
X1 + 2X2 <= 40
E-
3X1 + X2 >= 30
4X1 + 3X2 >= 60
X1, X2 >= 0 (Ans. X1=6, X2=12, Z=240)
(f) Min. Z = 30 X1 + 15 X2
CC

subject to:
5X1 + X2 >= 10
X1 + X2 >= 6
X1 + 4X2 >= 12
X1, X2 >=0 (Ans. X1=1, X2= 5, Min. Z = 105)
(g) Min. Z = 12 X1 + 15 X2

subject to:
X1 <= 5
X2 >= 3
(c)

X 1 + X2 = 6
X1, X2 >= 0 (Ans. X1 =3, X2 = 3, Z = 81)
 Quantitative Techniques for Management Applications

182
11. Solve the following problems by simplex method:

S
Notes
Max. Z = 2X1 + 4X2
___________________
Subject to:
___________________
2X1 + 3X2 <= 48
___________________

E
X1 + 3X2 <= 42
___________________
X1 + X2 <= 21
___________________
X1, X2 >= 0 (Ans. X1=6, X2=12, Z=60)
___________________

UP
___________________
12 Max. Z = 4X1 + 10X2

___________________ Subject to:

___________________ 2X1 + X2 <= 50

___________________ 2X1 + 5X2 <= 100

2X1 + 3X2 <= 90
X1 X2 >= 0 (Ans. X1=0, X2=20, Z=200)
13. Min. Z = 3X1 + 2X2+ X3
E-
Subject to:
-3X1 + 2X2 + 2X3 = 8
-3X1 + 4X2 + X3 = 7
X1, X2, X3 >=0

(Ans. Unbounded Solution, Hint: all the elements in key column

CC

have negative sign and Cj-Zj >0)

(Ans. 4 insertions in TV & 0 in FM radio)

14. Obtain the dual of the following:
Max. Z = -3X1 – 2X2

subject to:
X1 + X2 >= 1
X1 +2X2 >= 10
(c)

X1 + X2 <= 7
X2 <= 3
and X1, X2 >=0
UNIT 12: Linear Programming

15. Find out the dual of the primal given below: 183

S
Notes
Min. Z = 10 W1 + 8 W2
___________________
subject to:
___________________
W1 + W2 >= 1

E
___________________
W1 + 3W2 >= 4
___________________
2W1 - W2 <= 12
___________________
and W1, W2 >=0
___________________

UP
16. Write the dual of the following primal problem:
___________________
Max. Z = 5W1 + 6 W2
___________________
subject to: ___________________
W1 + 2W2 = 5 ___________________
-W1 + 5W2 >= 3
4W1 + 7W2 <= 8
W1 unrestricted in sign and W2 >=0
E-
Further Readings

Books
R S Bhardwaj, Mathematics for Economics and Business, Excel
Books, New Delhi, 2005
CC

D C Sanchethi and V K Kapoor, Business Mathematics

Sivayya and Sathya Rao, an Introduction to Business Mathematics

Web Readings
http://www.math.ucla.edu/~tom/LP.pdf

www.managementstudyguide.com

www.textbooksonline.tn.nic.in

www.mathbusiness.com
(c)
 Quantitative Techniques for Management Applications

184

S
Notes

___________________

___________________

___________________

E
___________________

___________________

___________________

UP
___________________

___________________

___________________

___________________
E-
CC
(c)
UNIT 13: Transportation Models

Unit 13
185

S
Activity
Notes
What is transportation

Transportation Models
___________________
models?
___________________

E
___________________
Objectives
___________________
After completion of this unit, the students will be aware of the following
topics: ___________________

\ The significance of transportation models ___________________

UP
\ The types of transportation models ___________________
\ The applications of transportation models
___________________
\ How to solve the transportation problems using IFS and optimality test
___________________

___________________
Introduction
The TRANSPORTATION model of linear programming is a flow
optimization technique. It is a special case that is somewhat easier
to solve than the general L.P. model. It is used to produce optimal
E-
assignments of origin quantities to destinations. A condition is that
the sum of the origin quantities must equal the destination
demands. F.L. Hitchcock contributed significantly in developing
transportation models. Since transportation is an economic activity
all the various accounting, financial, economic, & econometric
models are used. Much of the transport activity that uses ‘civil
CC

engineering’ is involved with ‘public works’. Economic and

accounting models for private and public enterprises are generally
used to assist major decisions & policy formulation.
The transportation problem is a special case of Linear
Programming which is concerned with the distribution of a certain
product/commodity from a number of sources (origins) to the
number of destinations. It can be tabulated in a matrix called
transportation matrix. In this matrix, each row denotes sources
(factories), each column denotes the destination (warehouses) and
each cell shows the cost of transportation per unit (Cij) from ith
factory to jth warehouse. Ai and Bj denote the supply and demand
(c)

respectively.

Factory/ D1 D2 D3 D4 Supply
Warehouse
O1 C11 C12 C13 C14 A1
Contd…
 Quantitative Techniques for Management Applications

186 O2 C21 C22 C23 C24 A2

S
Notes O3 C31 C32 C33 C34 A3

___________________ O4 C41 C42 C43 C44 A4

Demand B1 B2 B3 B4 Total
___________________

___________________ Objective: The objective of transportation problem is to minimize

E
the cost of transportation under the given supply and demand
___________________
constraints.
___________________

___________________
Check Your Progress

UP
Fill in the blanks:
___________________

___________________
1. The ________________ model of linear programming is a
flow optimization technique.
___________________
2. The objective of transportation problem is to minimize
___________________
the cost of transportation under the given supply and
________________.

Demand, Transport Demand and Market Models

E-
“At any one price, there is some quantity of a product which an
individual consumer is willing and able to purchase over a given
period of time. If the price changes, the quantity purchased will
also change. Economists call the relationship between the price of
a commodity and the quantity purchased during some specified
period of time the demand for that commodity.
CC

When a consumer is willing and able to purchase some quantity of

a commodity at the existing market price, he is said to have an
effective demand for that good. This means that the buyer has the:

1. desire to make a purchase

2. willingness to pay the price

3. ability to pay the price

Transport demand is usually categorized as:

(c)

z Commodity or Goods movements, i.e. quantified by mass,

weight, volume or number of items.

z Person Travelling, Passengers, Passenger Trips, People

Movements, etc.
UNIT 13: Transportation Models

Demand or Market Models tend to be disaggregated by trip 187

S
purpose, or commodity. These models are either: Notes

z Cross Section, i.e. they quantify the movements for a short ___________________

time period for a number of Origins and Destinations. ___________________

Time Series, i.e. they represent one movement over a number

E
z ___________________
of time periods. ___________________

z General, i.e. they attempt to combine cross section and time ___________________
series aspects much like space-time representations.
___________________

UP
The models may focus on a single market segment or several. Some ___________________
further categorization:
___________________
z Aggregate, i.e. the producing activity is treated as single ___________________
strata
___________________
z Disaggregate which subdivides the activity into a number of
strata

z Multimodal where a number of competing and

complimentary modes are used to satisfy a market
E-
z Abstract Mode, an econometric approach that is useful in
multimodal models

Applications
The common uses are for planning, design, operations and
management. Each of the functions requires that the models
CC

reflect behaviour of the chosen system and scenario. Each of the

comments below applies equally to the four activities mentioned
above. However the costs and difficulties of complicated modelling
should be matched with the expected payoff from the exercise.

1. Identify technically efficient solutions to transport resource

allocation: The models can be run as:

(a) Transport Costing Models (also net revenue)

(b) Transport Production Models

(c) Resource Costing Models (also net revenue)

(c)

(d) Resource Production Models

2. They can be used to OPTIMIZE the system for a given level of

production for a given amount of service. This can lead to
development of cost effective relationships with different
 Quantitative Techniques for Management Applications

188
levels of production with fixed levels of plant, labour or

S
Notes
Activity investment.
What are the different types of
___________________
transportation problems? 3. Give the ability to model as described above the domain can
___________________ be increased to simulate performance with some of the inputs
___________________ as random variables, and for different scenarios.

E
___________________ Check Your Progress
___________________ Fill in the blanks:
___________________ 1. Cross Section quantifies the movements for a _______

UP
___________________ time period for a number of Origins and Destinations.

___________________ 2. Time Series represent one movement over a

___________________
____________ time periods.

___________________
Types of Transportation Problem
(i) Minimization (Cost or Distance)

(ii) Maximization (Return or Revenue)

E-
(iii) Balanced (Supply = Demand)

(iv) Unbalanced (Supply not equal to Demand)

(v) Restricted

(vi) Transshipment
CC

Balanced Transportation Problem

When the total supplies of all the sources are equal to the total
demand of all destinations, the problem is a balanced
transportation problem.

Total supply = Total demand

m n

∑a = ∑b
i =1
i
j =1
j

Example 13.1
(c)

Consider the following transportation problem and develop a linear

programming (LP) model.
UNIT 13: Transportation Models

Table 13.1: Transportation Problem 189

S
Source Destination Notes
1 2 3 Supply ___________________
1 15 20 30 350
___________________
2 10 9 15 200

E
3 14 12 18 400 ___________________
Demand 250 400 300 ___________________

Solution: Let xij be the number of units to be transported from the ___________________

source i to the destination j, where i = 1, 2, 3,…m and j = 1, 2, ___________________

UP
3,…n.
___________________
The linear programming model is ___________________
Minimize Z = 15x11+20x12+30x13+10x21+9x22+15x23+14x31+12x32+18x33 ___________________

Subject to constraints, ___________________

x11+x12+x13 ≤ 350 (i)

x21+x22+x23 ≤ 200 (ii)

x31+x32+x33 ≤ 400 (iii)

E-
x11+ x12+x31 = 250 (iv)

x12+x22+x32 = 400 (v)

x13+x23+x33 = 300 (vi)

xij ≥ 0 for all i and j.
CC

In the above LP problem, there are m × n = 3 × 3 = 9 decision

variables and m + n = 3 + 3 = 6 constraints.

Unbalanced Transportation Problem

When the total supply of all the sources is not equal to the total
demand of all destinations, the problem is an unbalanced
transportation problem.

Total supply ≠ Total demand

m n

∑ ai ≠ ∑b j
(c)

i =1 j =1

Demand Less than Supply

In real-life, supply and demand requirements will rarely be equal.
This is because of variation in production from the supplier end,
and variations in forecast from the customer end. Supply
 Quantitative Techniques for Management Applications

190 variations may be because of shortage of raw materials, labour

S
Notes problems, improper planning and scheduling. Demand variations
___________________ may be because of change in customer preference, change in prices
and introduction of new products by competitors. These
___________________
unbalanced problems can be easily solved by introducing dummy
___________________

E
sources and dummy destinations. If the total supply is greater
___________________ than the total demand, a dummy destination (dummy column)
___________________ with demand equal to the supply surplus is added. If the total
demand is greater than the total supply, a dummy source (dummy
___________________
row) with supply equal to the demand surplus is added. The unit

UP
___________________ transportation cost for the dummy column and dummy row are
___________________ assigned zero values, because no shipment is actually made in case
___________________
of a dummy source and dummy destination.

___________________ Example 13.2: Check whether the given transportation problem

shown in Table is a balanced one. If not, convert the unbalanced
problem into a balanced transportation problem.

Table 13.2: Transportation Model with

Supply Exceeding Demand
E-
Source Destination Supply
1 2 3
1 25 45 10 200
2 30 65 15 100
3 15 40 55 400
Demand 200 100 300
CC

Solution: For the given problem, the total supply is not equal to
the total demand.
3 3

∑a i =1
i ≠ ∑b
j =1
j

3 3
since, ∑ ai = 700 and ∑ b j = 600
i =1 j =1

The given problem is an unbalanced transportation problem. To

convert the unbalanced transportation problem into a balanced
problem, add a dummy destination (dummy column).
(c)

i.e., the demand of the dummy destination is equal to,

3 3

∑ ai −
i =1
∑b
j =1
j
UNIT 13: Transportation Models

191
Thus, a dummy destination is added to the table, with a demand of

S
Notes
100 units. The modified table is shown in Table which has been
converted into a balanced transportation table. The unit costs of ___________________
transportation of dummy destinations are assigned as zero. ___________________

Table 13.3: Dummy Destination Added

E
___________________

Source Destination Supply ___________________

1 2 3 4 ___________________

1 25 45 10 0 200 ___________________

UP
2 30 65 15 0 100 ___________________
3 15 40 55 0 400 ___________________
Demand 200 100 300 100 700/700 ___________________

Similarly, ___________________

n m
If ∑ b j > ∑ ai then include a dummy source to supply the excess
j =1 i =1

demand.
E-
Demand Greater than Supply

Example 13.3: Convert the transportation problem shown in table

into a balanced problem.
Table 13.4: Demand Exceeding Supply

Source Destination Supply

CC

1 2 3 4

1 10 16 9 12 200

2 12 12 13 5 300

3 14 8 13 4 300

Demand 100 200 450 250 1000/800

Solution: The given problem is,

4 3

∑b > ∑ a
j =1
j
i =1
j
(c)

3 4

∑ ai = 800 and
i =1
∑b
j =1
j = 1000

The given problem is an unbalanced one. To convert it into a

balanced transportation problem, include a dummy source (dummy
row) as shown in the table given below.
 Quantitative Techniques for Management Applications

192
Table 13.5: Balanced TP Model

S
Notes
Source Destination Supply
___________________ 1 2 3 4
___________________ 1 10 16 9 12 200
2 12 12 13 5 300
___________________

E
3 14 8 13 4 300
___________________
Demand 100 200 450 250 1000/800
___________________
Rim Condition: This is the necessary and sufficient condition for
___________________ determining the optimal solution any transportation problem. The

UP
___________________
condition is:
Total Supply = Total Demand
___________________
a i = bj
___________________

___________________ Check Your Progress

Fill in the blanks
1. ________ is the necessary and sufficient condition for
determining the optimal solution any transportation
E-
problem.
2. In real-life, supply and demand requirements will
________ be equal.

Degeneracy in the Transportation Problem

The degeneracy occurs in the transportation problem when we find
CC

IFS and observe that if the number of occupied cells is less than
the total number of rows plus columns minus one). Thus the
formula becomes:
Co < m + n – 1

Where Co = No. of occupied cells

m = No. of rows
n = No. of columns

Methods of Transportation Problem

The steps for solving the transportation problem:
1. Formulate the problem and set up in the matrix form.
(c)

2. Obtain IFS by applying any of the methods.

3. Apply the optimality test (MODI or STEPPING STONE

METHOD) for obtaining the optimal solution.
UNIT 13: Transportation Models

The transportation problem can be solved in two phases. In the 193

S
first phase, IFS (Initial Feasible Solution) is calculated and in the Notes
second phase, optimal solution is calculated. ___________________

For obtaining the IFS, the following methods are used: ___________________

E
(i) NWCM (North West Corner Method) ___________________

(ii) LCM/MMM (Least Cost Method/Matrix Minima Method) ___________________

___________________
(iii) VAM (Vogel’s Approximation Method)
___________________
For feasible solution, the methods used are:

UP
___________________
(i) MODI (Modified Distribution Method)
___________________
(ii) SSM (Stepping Stone Method)
___________________
Before applying IFS, check the rim condition (ai = bj), which must
___________________
be satisfied.

NWCM (North West Corner Method):

1. Choose the cell in north west corner.

E-
2. Find out minimum of supply and demand i.e. Min. (ai, bj).

3. Allocate min.(ai, bj) in the north west cell and exhaust the row
(column) if supply (demand) is satisfied and adjust the
balance.

4. Repeat the steps 1 onwards till all the supply and demand are
satisfied.
CC

LCM/MMM (Least Cost Method/Matrix Minima Method)

1. Choose the cell with minimum cost.
2. Find out minimum of supply and demand i.e. Min. (ai, bj).

3. Allocate min.(ai, bj) in the north west cell and exhaust the row
(column) if supply (demand) is satisfied and adjust the
balance.

4. Repeat the steps 1 onwards till all the supply and demand are
satisfied.
(c)

VAM (Vogel’s Approximation Method)

1. Calculate the penalty for each row or column by subtracting
the smallest element from the next smaller element.

2. Choose the highest penalty out of all the penalties and choose
the cell with minimum cost in the corresponding row (column).
 Quantitative Techniques for Management Applications

194
In case of tie between two highest penalties, choose

S
Notes
Activity arbitrarily.
What is optimality test?
___________________ 3. Find out minimum of supply and demand i.e. Min. (ai, bj)
___________________ corresponding to the chosen cell.
___________________ 4. Allocate min.(ai, bj) in the cell with minimum cost and exhaust

E
___________________ the row (column) if supply (demand) is satisfied and adjust the
balance.
___________________
5. Repeat the steps I onwards till all the supply and demand are
___________________
satisfied.

UP
___________________
Optimality Test
___________________

___________________
MODI Method:

___________________ Prior to applying this method, the following condition must be

satisfied:
Co = m+n-1

If this condition is not satisfied, degeneracy occurs. Degeneracy can

E-
be removed by putting Delta in the unoccupied cell having
minimum cost.

1. Construct a transportation table with given cost and

allocations as per IFS through any of these methods.

2. For occupied cells, calculate index numbers Ui and Vj for rows

CC

and columns respectively. Values of these index numbers are

calculated by:
Cij = Ui + Vj

3. Opportunity cost is computed for all the unoccupied cells by

using the following equation:
Dij = Cij – (Ui + Vj)

4. Examine unoccupied cells evaluation for opportunity cost (Dij);

(a) If Dij > 0, Cost of transportation will increase, the solution

(c)

is optimal.

(b) If Dij < 0, Cost of transportation will decrease, the solution

is not optimal.
UNIT 13: Transportation Models

195
(c) If Dij = 0, Cost of transportation will not change, alternate

S
Notes
solution exists.
___________________
In case of Dij < 0, loop is constructed for which following steps
___________________
are required:

E
5. Select an unoccupied cell with largest negative opportunity ___________________

cost. ___________________

6. Constructed a closed path for the unoccupied cell determined ___________________

in previous step and assign plus (+) and minus (-) sign
___________________
alternatively beginning with plus sign for the selected

UP
unoccupied cell. ___________________

7. Assign as many units as possible to the unoccupied cell ___________________

satisfying the rim condition. The smallest allocation in a cell ___________________
with –ve sign on the closed path indicates number of units
___________________
that can be shipped to the unoccupied cells. This quantity is
added to all the occupied cells on the path marked with +ve
sign.
8. Go to step 2 and iterate all the steps until all Dij become
E-
positive to reach the optimal solution. Then calculate the
transportation cost.

* The method discussed here is applicable for case of minimization.

In case of maximization, the problem is converted into
minimization by subtracting all the elements from the highest
element.
CC

Check Your Progress

Fill in the blanks:
1. The __________ occurs in the transportation problem
when we find IFS and observe that if the number of
occupied cells are less than the total number of rows
plus columns minus one.
2. In MODI optimality test C0 = m+ ________________.

Summary
The transportation problem is a special case of Linear
(c)

Programming which is concerned with the distribution of a certain

product/commodity from a number of sources (origins) to the
number of destinations. It can be tabulated in a matrix called
transportation matrix.
 Quantitative Techniques for Management Applications

196 The degeneracy occurs in the transportation problem when we find

S
Notes IFS and observe that if the numbers of occupied cells are less than
___________________ the total number of rows plus columns minus one).
___________________ The transportation problem can be solved in two phases. In the
___________________ first phase, IFS (Initial Feasible Solution) is calculated and in the

E
second phase, optimal solution is calculated.
___________________

___________________
Lesson End Activity
___________________
Illustrate the preparation of transportation matrix with a suitable

UP
___________________
example.
___________________

___________________ Keywords
___________________ Transportation Model: The transportation model of linear
programming is a flow optimization technique.
Effective Demand: When a consumer is willing and able to
purchase some quantity of a commodity at the existing market
E-
price, he is said to have an Effective Demand for that good.

Questions for Discussion

1. Define transportation problem.
2. What are the different types of transportation problems?
3. What are the key methods of obtaining IFS?
CC

4. An oil company has three refineries located in the country.

The daily oil production (in million tonnes) is as follows:
Refineries I II III
Oil produced 6 1 10

Each day, the refineries must satisfy the needs for their
distribution centres. Minimum requirement at each centre is
as under:
Distribution Centre I II III IV
Oil supply 7 5 3 2

The cost in thousands of rupees of shipping one million tonnes

(c)

from each refinery to each distribution centre is as per the

table given below:
Dist. Centres/Refineries D1 D2 D3 D4
R1 2 3 11 7
R2 1 0 6 1
R3 5 8 15 9
UNIT 13: Transportation Models

Determine initial feasible solution by a) NWCM b) LCM c) 197

S
VAM method. Notes

(Ans. `11,600, `11,200, `10,200) ___________________

5. A car hire company has one car at each of five depots A,B,C,D ___________________
and E. A customer requires a car in each town P,Q,R,S,T.

E
___________________
Distances (given in kms) between depots and towns are given
in the following table: ___________________

Depot/Town A B C D E ___________________
P 160 130 175 190 200 ___________________

UP
Q 135 120 130 160 175
___________________
R 140 110 155 150 185
S 150 50 80 80 110 ___________________
T 55 35 70 80 105 ___________________

(Ans. 570 kms) ___________________

Further Readings

Books
E-
R S Bhardwaj, Mathematics for Economics and Business, Excel
Books, New Delhi, 2005

D C Sanchethi and V K Kapoor, Business Mathematics

Sivayya and Sathya Rao, An Introduction to Business Mathematics

CC

Web Readings
http://www.math.ucla.edu/~tom/LP.pdf

www.managementstudyguide.com

www.textbooksonline.tn.nic.in

www.mathbusiness.com
(c)
 Quantitative Techniques for Management Applications

198

S
Notes

___________________

___________________

___________________

E
___________________

___________________

___________________

UP
___________________

___________________

___________________

___________________
E-
CC
(c)
UNIT 14: Assignment Problem

Unit 14
199

S
Activity
Notes
What is assignment problem?

Assignment Problem
___________________

___________________

E
___________________
Objectives
___________________
After completion of this unit, the students will be aware of the following
topics: ___________________

\ The objective of assignment models ___________________

UP
\ The applications of assignment models ___________________
\ The Hungarian method for solving the assignment problems
___________________

___________________
Introduction
___________________
An assignment problem is a special type of transportation problem.
The method to solve assignment problems was introduced by D.
Konig, a Hungarian mathematician. In such models, only one unit
can be supplied to each destination from each source, for example,
E-
to assign one job to each facility in order to achieve the minimum
possible cost.

Objective: The objective of assignment model is to assign a

number of resources to an equal number of activities so as to
minimize the cost or maximize the profit by optimal allocation.
CC

Applications
(i) Assignment of machines to jobs

(ii) Assignment of workers to various tasks

(iii) Assignment of sales representatives to sales territories

(iv) Assignment of contracts to bidders

(v) Assignment of buses/airlines/trains to various routes

(vi) Assignment of officers to various offices

Types of assignment problems:

(c)

(i) Minimization (Cost, time or Distance)

(ii) Maximization (Profit or Revenue)

(iii) Balanced (No. of rows = No. of columns)

(iv) Unbalanced (No. of rows /= No. of columns)

 Quantitative Techniques for Management Applications

200
(v) Restricted

S
Notes
Activity
What is Hungarian Method for
___________________ (vi) Crew assignment
Solving Assignment Problem?
___________________ Check Your Progress
___________________ Fill in the blanks:

E
___________________
1. An assignment problem is a special type of ___________
___________________ problem.
___________________ 2. The objective of assignment model is to assign a number

UP
___________________ of resources to an equal number of activities so as to
___________ the cost or maximize the profit by optimal
___________________
allocation
___________________

___________________
Hungarian Method for Solving Assignment Problem
Step 1: In a given problem, if the number of rows is not equal to
the number of columns and vice versa, then add a dummy
row or a dummy column. The assignment costs for dummy
E-
cells are always assigned as zero.

Step 2: Reduce the matrix by selecting the smallest element in

each row and subtract with other elements in that row.

Step 3: Reduce the new matrix column-wise using the same

method as given in step 2.

Step 4: Draw minimum number of lines to cover all zeros.

CC

Step 5: If Number of lines drawn = order of matrix, then

optimality is reached, so proceed to step 7. If optimality is
not reached, then go to step 6.

Step 6: Select the smallest element of the whole matrix, which is

NOT COVERED by lines. Subtract this smallest element
with all other remaining elements that are NOT
COVERED by lines and add the element at the
intersection of lines. Leave the elements covered by single
line as it is. Now go to step 4.

Step 7: Take any row or column which has a single zero and
(c)

assign by squaring it. Strike off the remaining zeros, if

any, in that row and column (X). Repeat the process until
all the assignments have been made.

Step 8: Write down the assignment results and find the minimum
cost/time.
UNIT 14: Assignment Problem

201
Note: While assigning, if there is no single zero exists in the row or

S
Notes
column, choose any one zero and assign it. Strike off the remaining
zeros in that column or row, and repeat the same for other ___________________
assignments also. If there is no single zero allocation, it means ___________________
multiple numbers of solutions exist. But the cost will remain the

E
___________________
same for different sets of allocations.
___________________
Example: Assign the four tasks to four operators. The assigning
___________________
costs are given in Table 14.1
___________________

UP
Table 14.1: Assignment Problem
___________________
Operators
___________________
1 2 3 4
___________________
A 20 28 19 13
___________________
Tasks B 15 30 31 28

C 40 21 20 17

D 21 28 26 12
E-
Solution:

Step 1: The given matrix is a square matrix and it is not

necessary to add a dummy row/column

Step 2: Reduce the matrix by selecting the smallest value in each

row and subtracting from other values in that
CC

corresponding row. In row A, the smallest value is 13, row

B is 15, row C is 17 and row D is 12. The row wise reduced
matrix is shown in Table 14.2:
Table 14.2: Row-wise Reduction
Operators

1 2 3 4

A 7 15 6 0

Tasks B 0 15 16 13

C 23 4 3 0
(c)

D 9 16 14 0

Step 3: Reduce the new matrix given in table by selecting the

smallest value in each column and subtract from other
values in that corresponding column. In column 1, the
 Quantitative Techniques for Management Applications

202
smallest value is 0, column 2 is 4, column 3 is 3 and

S
Notes column 4 is 0. The column-wise reduction matrix is shown
___________________ in Table 14.3.
___________________ Table 14.3: Column-wise Reduction Matrix
Operators
___________________

E
___________________ 1 2 3 4

___________________ A 7 11 3 6

___________________ Tasks B 0 11 13 13

UP
___________________ C 23 0 0 0

___________________ D 9 12 11 0

___________________
Step 4: Draw minimum number of lines possible to cover all the
___________________ zeros in the matrix given in Table 14.4
Table 14.4: Matrix with all Zeros Covered
Operators

1 2 3 4
E-
A 7 11 3 0
No. of lines
Tasks B 0 11 13 13 drawn
≠ order of
C 23 0 0 0 matrix

D 9 12 11 0
CC

The first line is drawn crossing row C covering three

zeros, second line is drawn crossing column 4 covering two
zeros and third line is drawn crossing column 1 (or row B)
covering a single zero.

Step 5: Check whether number of lines drawn is equal to the

order of the matrix, i.e., 3 ≠ 4. Therefore optimality is not
reached. Go to step 6.

Step 6: Take the smallest element of the matrix that is not

covered by single line, which is 3. Subtract 3 from all
other values that are not covered and add 3 at the
intersection of lines. Leave the values which are covered
(c)

by single line. Table 14.5 shows the details.

UNIT 14: Assignment Problem

Table 14.5: Subtracted or Added to Uncovered Values and Intersection 203

S
Lines Respectively
Notes
Operators
___________________
1 2 3 4
___________________
A 7 9 0 0

E
___________________
Tasks B 0 9 10 13
___________________
C 26 0 0 3 ___________________
D 9 9 8 0 ___________________

UP
Step 7: Now, draw minimum number of lines to cover all the zeros ___________________
and check for optimality. Here, in Table 14.6, minimum ___________________
number of lines drawn is 4, which is equal to the order of
matrix. Hence optimality is reached. ___________________

Table 14.6: Optimality Matrix ___________________

Operators

1 2 3 4

A 7 9 0 0
E-
No. of lines
Tasks B 0 9 10 13
drawn =
order of
C 26 0 0 3
matrix

D 9 9 8 0

Step 8: Assign the tasks to the operators. Select a row that has a
single zero and assign by squaring it. Strike off remaining
CC

zeros if any in that row or column. Repeat the assignment

for other tasks. The final assignment is shown in Table
14.7.
Table 14.7: Final Assignment
Operators

1 2 3 4

A 7 9 0
×
0
Tasks B 0 9 10 13

C 26 0 0
× 3
(c)

D 9 9 8 0
 Quantitative Techniques for Management Applications

204
Therefore, optimal assignment is:

S
Notes

___________________ Task Operator Cost

___________________ A 3 19

___________________ B 1 15

E
___________________ C 2 21

___________________ D 4 12

Total Cost = ` 67.00

___________________

UP
___________________ Check Your Progress
___________________ Fill in the blanks
___________________ 1. While assigning, if there is no single zero exists in the
___________________ row or column, choose _______zero and assign it.
2. If there is no single zero allocation, it means
_____________ of solutions exist

Summary
E-
An assignment problem is a special type of transportation problem.
The objective of assignment model is to assign a number of
resources to an equal number of activities so as to minimize the
cost or maximize the profit by optimal allocation.
CC

Lesson End Activity

Study the application areas where Hungarian method can be
beneficial.

Keywords

Assignment Problem: An assignment problem is a special type of

transportation problem.
Objective of Assignment Model: Objective of assignment model
is to assign a number of resources to an equal number of activities
(c)

Questions for Discussion

1. Define assignment model.

2. What is the objective and application of assignment model?

UNIT 14: Assignment Problem

205
3. What are the different types of assignment problems?

S
Notes
4. Write a note on Hungarian method. ___________________

5. Due to absence of a workman, an officer has to assign four out ___________________

of five different jobs to four workers with the performance

E
___________________
(cost) matrix given below, determine an optimal solution.
___________________
Worker/Jobs A B C D
1 3 6 5 3 ___________________

2 4 9 3 2 ___________________

UP
3 11 2 4 6
___________________
4 10 4 6 5
5 11 12 14 10 ___________________

___________________
(Ans. ` 13,000)
___________________
6. A company is faced with the problem of assigning five
different machines to five different jobs with a view to
minimize total cost. The costs (in thousands) are estimated
and shown in the following table:
E-
Jobs
Machines 1 2 3 4 5
I 2.5 5 1 6 2
II 2 5 1.5 7 3
III 3 6.5 2 8 3
IV 3.5 7 3 9 4.5
CC

V 4 7 3 9 6

(Ans. ` 20,000)

Further Readings

Books
R S Bhardwaj, Mathematics for Economics and Business, Excel
Books, New Delhi, 2005
D C Sanchethi and V K Kapoor, Business Mathematics
Sivayya and Sathya Rao, An Introduction to Business Mathematics
(c)

Web Readings
http://www.math.ucla.edu/~tom/LP.pdf
www.managementstudyguide.com
www.textbooksonline.tn.nic.in
www.mathbusiness.com
 Quantitative Techniques for Management Applications

206

S
Notes

___________________

___________________

___________________

E
___________________

___________________

___________________

UP
___________________

___________________

___________________

___________________
E-
CC
(c)
UNIT 15: Case Studies

Unit 15
207

S
Notes

Case Studies
___________________

___________________

E
___________________
Objectives
___________________
After analyzing these cases, the student will have an appreciation of the
concept of topics studied in this Block. ___________________

___________________

UP
___________________
Case Study 1: Constructing Payoff Table
___________________
A company wants to increase its production beyond its existing
capacity. It has finally arrived at two approaches to increase the ___________________
capacity (1) Expansion, at a cost of ` 80 Lakh or (2) Modernization
___________________
at a cost of ` 50 Lakh. Both approaches would require 8 months
for implementation. The Board of Directors feels that during
implementation and thereafter the demand will either be very
high or moderate. The probability for very high demand is
estimated as 0.35 and for moderate it is 0.65. If demand is very
E-
high, expansion would result additional profit of 120 Lakh, but on
the other hand modernization would bring additional 60 Lakh
only. It is estimated that when demand is moderate, the
comparable profit would be 70 Lakh and 50 Lakh for
modernization.

Questions
CC

(a) Construct the Payoff (Profit) Table.

(b) What is the optimum strategy for the company?

(c) Calculate EMV and EVPI.

(d) Calculate EOL.

Source: MBCQ-721D_Quantitative Techniques for Management Applications_CCE_UPES
(c)
 Quantitative Techniques for Management Applications

208
Case Study 2: Media Mix Determination

S
Notes
An advertising agency wants to finalise its campaign and it plans
___________________ to target two types of audiences: customers with cell phone (Type
A) and customers not having cell phones (Type B). The total ad
___________________
budget is ` 2 lakhs. One insertion of TV ad on movie channel costs
___________________ ` 50,000 and one insertion on FM radio costs ` 20,000. As per

E
___________________ agreement, at least three insertions must be there for movie
channel and it can not exceed five in number. As per the findings
___________________
of research agency, a single TV ad reaches 350,000 customers in
___________________ target audience A and 150,000 in target audience B. One FM

UP
radio ad reaches 10,000 in target audience A and 90,000 in target
___________________
audience B.
___________________
Question
___________________
Determine the media mix to maximize the total reach. (Formulate
___________________ the problem and use simplex method)
Source: MBCQ-721D_Quantitative Techniques for Management Applications_CCE_UPES
E-
CC
(c)
UNIT 15: Case Studies

Case Study 3: Determining Optimal Solution 209

S
A company has plants P1, P2, P3 which supply to distribution Notes
centres D1, D2 and D3. Monthly factory capacities are 200, 160 and ___________________
90 units respectively whereas requirement of distribution centres
___________________
are 180, 120 and 150 respectively. Unit transportation cost is
given in the following table:

E
___________________

Distribution D1 D2 D3 ___________________
Centres\Plants
___________________
P1 16 20 12
P2 14 8 18 ___________________

UP
P3 26 24 16
___________________

Question ___________________

Determine the optimal solution for the company. ___________________

Source: MBCQ-721D_Quantitative Techniques for Management Applications_CCE_UPES ___________________
E-
CC
(c)
 Quantitative Techniques for Management Applications

210
Case Study 4: Determining Optimal Assignment

S
Notes
A trip from Delhi to Ajmer takes six hours by bus. The manager
___________________ has designed the time table of bus service both ways which is as
under:
___________________
Departure Route Arrival Arrival Route Departure
___________________

E
from Delhi No. at at Delhi No. from Ajmer
Ajmer
___________________ 06.00 I 12.00 11.30 a 05.30
07.30 II 13.30 15.00 b 09.00
___________________
11.30 III 17.30 21.00 c 15.00
19.00 IV 01.00 00.30 d 18.30
___________________
00.30 V 06.30 06.30 e 00.00

UP
___________________
The cost of providing service by the travel agency depends upon
___________________
the time spent by the bus crew away from their places in addition
___________________ to service time. There is a constraint that every crew should be
provided with at least 4 hours of rest before the return trip and at
___________________
the most he can stay for 24 hours before he goes for the return
trip. The crews are residing at rest house hired by travel agency
in Delhi as well as Ajmer. The manager wants to minimize the
waiting time of crews.
E-
Question

Determine the optimal assignment for the crews for various

routes. (There are five crews and five routes)
Source: MBCQ-721D_Quantitative Techniques for Management Applications_CCE_UPES
CC
(c)
UNIT 16: Game Theory

211

S
Notes

___________________

___________________

E
___________________

___________________

___________________

___________________

UP
___________________

___________________

___________________

___________________

BLOCK-IV
E-
CC
(c)
Detailed Contents Quantitative Techniques for Management Applications

212

S
Notes
UNIT 16: GAME THEORY
___________________ UNIT 18: DATA COLLECTION
z Introduction z Introduction
___________________
z Game z Data Characteristics
___________________

E
z Assumptions z Requirements of Data in the Organization
___________________
z Limitations z Problems and Limitations of Data Collection
___________________
z Two-person Zero-sum Game z Basic Classification of Data

z
___________________
Rules (Principles) of Dominance z Primary Data

UP
___________________ z Methods of Primary Data Collection
UNIT 17: MARKOVIAN MODEL
___________________ z Secondary Data
z Introduction
___________________
z Properties of Markov Models UNIT 19: PRESENTATION OF DATA
___________________
z Applications z Introduction

z Types z Categorical data and numeric data

UNIT 20: CASE STUDIES

E-
CC
(c)
UNIT 16: Game Theory

Unit 16
213

S
Activity
Notes
What is theory of game?

Game Theory
___________________

___________________

E
___________________
Objectives
___________________
After completion of this unit, the students will be aware of the following
topics: ___________________

\ The game theory and its assumptions ___________________

UP
\ The types, limitations and rules of game theory ___________________
\ The value of game using various methods
___________________

___________________
Introduction
___________________
Professor John Von Neumann and Oscar Morgenstern published
their book entitled “The Theory of Games and Economic
Behaviour” wherein they provided a new approach to many
problems involving conflicting situations. The theory of games
E-
attempts to provide the rational decision in the confronting
situations.

Game
The term “game” represents a conflict between two or more
individuals or groups or organizations. It is the science of conflict.
CC

It is applicable to those competitive situations which are

technically known as “competitive games”. The objective of game
theory is to determine the rules of rational behaviour in situations
wherein the outcome resulting from a decision made by one
individual depends not only on that individual’s choice but also on
the course of action taken by other interested individuals.

Assumptions
1. The players act rationally and intelligently.

2. The players attempt to maximise gains and minimise losses.

(c)

3. All relevant information is known to each player.

4. Each player has available to him a finite set of possible

courses of action.
 Quantitative Techniques for Management Applications

214
5. The players make individual decisions without direct

S
Notes communication.
___________________
6. The players simultaneously select their respective course of
___________________ action.
___________________
7. The payoff is fixed in advance.

E
___________________
Types of Games
___________________
1. Two-person games and n-person games
___________________

UP
___________________ 2. Zero sum and non-zero sum games

___________________ 3. Games of perfect information and imperfect information

___________________ 4. Games with finite number of moves
___________________
5. Cooperative and non-cooperative games

6. 2 × 2 two-person games, 2 × m and m × 2 games

7. 3 × 3 and larger games

E-
8. Constant sum games

Limitations
z Businessmen do not have the knowledge of game theory and
all the alternative strategies available to them or their
competitors.
CC

z The business environment is turbulent and there is a lot of

uncertainty. Thus game theory may not be giving accurate
results in such cases as outcome of a strategy may not be
known with certainty.

z The game theory may not be suitable for oligopoly situations

where there are number of companies/firms involved.

z Larger size games are very complicated and cannot be solved

manually.

z The assumption in game theory that one player tries to

maximise the gains and the other tries to minimise the losses
(c)

may not be true in case of today’s dynamic businessman.

Strategy: The strategy for a player is the list of all possible

courses of actions that he will take for every pay-off that might
arise.
UNIT 16: Game Theory

Pure Strategy: It is the decision rule which is always followed by 215

S
the player to select the particular course of action. Notes
Activity
How___________________
will you account that
Mixed Strategy: When the player has alternative courses of
game is a competitive
action and he has to select combination of these with some fixed ___________________
situation?
probabilities.

E
___________________

Check Your Progress ___________________

Fill in the blanks: ___________________

1. The ____________ for a player is the list of all possible ___________________

UP
courses of actions that he will take for every pay-off that ___________________
might arise.
___________________
2. ____________ theory may not be giving accurate results
___________________
in many cases.
___________________

Two-person Zero-sum Game

In a game with two players, if the gain of one player is equal to the
loss of another player, then the game is a two person zero-sum
E-
game.

A game in a competitive situation possesses the following

properties:

1. The number of players is finite.

2. Each player has finite list of courses of action or strategy.

CC

3. A game is played when each player chooses a course of action

(strategy) out of the available strategies. No player is aware of
his opponent's choice until he decides his own.

4. The outcome of the play depends on every combination of

courses of action. Each outcome determines the gain or loss of
each player.

Pure Strategies: Game with Saddle Point

The aim of the game is to determine how the players must select
their respective strategies such that the pay-off is optimized. This
(c)

decision-making is referred to as the minimax-maximin principle

to obtain the best possible selection of a strategy for the players.
In a pay-off matrix, the minimum value in each row represents the
minimum gain for player A. Player A will select the strategy that
gives him the maximum gain among the row minimum values. The
 Quantitative Techniques for Management Applications

216 selection of strategy by player A is based on maximin principle.

S
Notes Similarly, the same pay-off is a loss for player B. The maximum
___________________ value in each column represents the maximum loss for Player B.
Player B will select the strategy that gives him the minimum loss
___________________
among the column maximum values. The selection of strategy by
___________________

E
player B is based on minimax principle. If the maximin value is
___________________ equal to minimax value, the game has a saddle point (i.e.,
___________________ equilibrium point). Thus the strategy selected by player A and
player B are optimal.
___________________

UP
___________________ Example 16.1: Consider the example to solve the game whose pay-
off matrix is given in table as follows:
___________________
Table 16.1: Game Problem
___________________
Player B
___________________ 1 2
1 1 3
Player A
2 -1 6

The game is worked out using minimax procedure. Find the

E-
smallest value in each row and select the largest value of these
values. Next, find the largest value in each column and select the
smallest of these numbers. The procedure is shown in table below.

Table 16.2: Minimax Procedure

Player B

1 2 Row Min
CC

1 1 3 1
Player A
2 -1 6 -1

Col Max 1 6

If Maximum value in row is equal to the minimum value in

column, then saddle point exists.

Max Min = Min Max

1=1
Therefore, there is a saddle point.
(c)

The strategies are,

Player A plays Strategy A1, (A → A1).
Player B plays Strategy B1, (B → B1).

Value of game = 1.
UNIT 16: Game Theory

217
Mixed Strategies: Games without Saddle Point

S
Notes
For any given pay off matrix without saddle point the optimum
___________________
mixed strategies are shown in Table.
___________________
Table 16.3: Mixed Strategies

E
___________________
Player B
___________________
B1 B2
A1 a11 a12
___________________

Player A ___________________

UP
A2 a21 a22
___________________
Let p1 and p2 be the probability for Player A. ___________________

Let q1 and q2 be the probability for Player B. ___________________

Let the optimal strategy be SA for player A and SB for player B. ___________________

Then the optimal strategies are given in tables given below.

Table 16.4: Optimum Strategies
E-
A1 A2 B1 B2

SA = and SB =
p1 p2 q1 q2

p1 and p2 are determined by using the formulae,

a 22 –a 21
p1 = and p2 = 1 – p1
( a11 + a 22 ) – ( a12 + a 21 )
CC

a 22 –a12
q1 = and q2 = 1 – q 1
( a11 + a 22 ) – ( a12 + a 21 )
and the value of the game w.r.t player A is given by,
a11 a22 – a12a21

Value of the game, v = (a11+a22) – (a12+a21)

Check Your Progress

Fill in the blanks
(c)

1. The game is worked out using _______ procedure.

2. In a _________ matrix, the minimum value in each row
represents the minimum gain for a player.
 Quantitative Techniques for Management Applications

218
Rules (Principles) of Dominance

S
Notes
Activity
What are the principles of
___________________ Row Dominance: When each element in a row are less than or
dominance? equal to the corresponding element in another row, this row is
___________________
dominated and hence can be deleted from the payoff matrix.
___________________

E
Column Dominance: When each element in a column are less
___________________
than or equal to the corresponding element in another column, this
___________________ column is dominated and hence can be deleted from the payoff
matrix.
___________________

UP
___________________ Average Dominance: A strategy can also be dominated when it is
inferior to an average of two or more pure strategies.
___________________

___________________
Value of a game: It is the average payoff per play of game over an
extended of time.
___________________
Saddle Point: In a payoff matrix saddle point is one which is the
smallest value in its row and the largest value in the column. In
other words, it is the point where maximin is equal to minimax.

Example 16.2: Determine the value of the following payoff matrix:

E-
Player B
Player A [2 3]
[5 4]

Solution:

Player B
CC

Player A [2 3] Row min. 2 Maximin = 4

[5 4] Row min. 4
Col. Max. 5 4 Minimax = 4

Thus in this game, saddle point is 4. This is value of the game.

Such game is known as pure strategy game.
(c)
UNIT 16: Game Theory

Flow Chart of Game Theory Approach 219

S
Notes

___________________
Formulate the pay – off matrix
___________________

E
Apply maximin or manimax principle ___________________

___________________

Identify the value of game and Yes Is ___________________

write the optimal strategy there a saddle
of the players point?
___________________

UP
No ___________________

Is it ___________________
Solve by using algebraic or matrix Yes a 2 x 2 pay-off
method matrix
for mixed strategic games ___________________
game?
___________________
No

Use dominance rule to reduce the size

of the pay-off matrix to either 2 x 2.
2 x n or m x 2 size (order)

No
E-
Is
pay-off matrix
Yes reduced
to a 2 x 2
size?

No

Is
pay-off matrix
CC

Use graphical method to solve the Yes reduced to a

problem 2 x n or m x 2
size?

No

Formulate and solve as an LP problem

Figure 16.1: Game Theory Approach

Methods for Solving Games

(i) Short cut method

(ii) Graphical method

(c)

(iii) Algebraic method

(iv) Linear Programming method

 Quantitative Techniques for Management Applications

220
Check Your Progress

S
Notes
Fill in the blanks:
___________________

___________________
1. When each element in a column are ________ or
________ the corresponding element in another column.
___________________

E
2. In a payoff matrix ____________ is one which is the
___________________
smallest value in its row and the largest value in the
___________________ column.
___________________

UP
___________________ Summary
___________________ The term "game" represents a conflict between two or more
___________________ individuals or groups or organizations. It is the science of conflict.
It is applicable to those competitive situations which are
___________________
technically known as "competitive games". The objective of game
theory is to determine the rules of rational behaviour in situations
wherein the outcome resulting from a decision made by one
individual depends not only on that individual's choice but also on
E-
the course of action taken by other interested individuals.
The key methods of solving games are:
(i) Short cut method
(ii) Graphical method
(iii) Algebraic method
(iv) Linear Programming method
CC

Lesson End Activity

Determine the value of the game and optimum strategies for the
following matrix:

Player B
−3 −4
Player A −6 −2
−7 −1
−2 −9
(c)

Keywords
Strategy: The strategy for a player is the list of all possible
courses of actions that he will take for every pay-off that might
arise.
UNIT 16: Game Theory

221
Pure Strategy: It is the decision rule which is always followed by

S
Notes
the player to select the particular course of action.
___________________
Mixed Strategy: When the player has alternative courses of
action and he has to select combination of these with some fixed ___________________

probabilities.

E
___________________

Saddle Point: In a payoff matrix saddle point is one which is the ___________________
smallest value in its row and the largest value in the column. ___________________

Questions for Discussion ___________________

UP
___________________
1. Define game and its types.
___________________
2. Define the following:
___________________
(a) Saddle Point ___________________

(b) Pure and Mixed Strategies

(c) Zero-sum game

3. What are the assumptions made in the theory of games?

E-
4. Describe rules of dominance with examples.

5. What are the various methods for solving a game theory and
their suitability?

6. Find out the value of the game, the payoff (in rupees) is given
in the following matrix:
CC

Player A / Player B Strategy I Strategy II

Strategy I 10 14
Strategy II 7 12

7. The conditional gains to the workers association (in

thousands) against management strategies are given in the
following payoff table:
Association Management Strategies
Strategies M1 M2 M3 M4
A1 20 15 12 35
A2 25 14 8 10
(c)

A3 40 2 10 5
A4 -5 4 11 0

8. Two petroleum companies X and Y are competing for business.

The matrix shows the gains to the company X (Assume the
game is zero sum).
 Quantitative Techniques for Management Applications

222 Company X’s Gain

S
Notes X/Y Sales Promotion Advertising Exhibition
___________________ Sales Promotion 60 50 40
Advertising 70 70 50
___________________
Exhibition 80 60 75
___________________
Determine the optimal strategies and value of the game.

E
___________________ 9. Determine the value of the following game:
___________________ Player Y
I II III IV
___________________
I 3 2 4 0

UP
___________________ II 3 4 2 4
___________________ Player X
III 4 2 4 0
___________________
IV 0 4 0 8
___________________
10. The payoff matrix is given below. Determine the optimum
strategies and value of the game. (Solve graphically)
Player B
3 6 -1 0 -3
E-
Player A
2 3 -4 2 -1

Further Readings

Books
CC

R S Bhardwaj, Mathematics for Economics and Business, Excel

Books, New Delhi, 2005

D C Sanchethi and V K Kapoor, Business Mathematics

Sivayya and Sathya Rao, An Introduction to Business Mathematics

Web Readings
http://www.math.ucla.edu

www.textbooksonline.tn.nic.in

www.mathbusiness.com
(c)
UNIT 17: Markovian Model

Unit 17
223

S
Activity
Notes
What is markovian model?

Markovian Model
___________________

___________________

E
___________________
Objectives
___________________
After completion of this unit, the students will be aware of the following
topics: ___________________

\ The properties and applications of Markovian model ___________________

UP
\ The types of Markov process and various states involved ___________________
\ How to determine the future market share and market share in steady
state condition ___________________

___________________

Introduction ___________________

Markov models are used to analyze states of stochastic system to

describe its position at any instant of time.
E-
Properties of Markov Models
(i) It is a stochastic (probabilistic) process.

(ii) Markov Process is a sequence of experiments in which each

experiment has certain possible outcomes.

(iii) There is a finite set of states.

CC

(iv) The process can be only in one state at a given time.

(v) The probability of moving from one state to another or

remaining in the same state in a single time period is called
transition probability (Pij). It always remains constant and
0 <= Pij <= 1

(vi) The probability of transition from a given state to future state

is dependent on the present state.

Check Your Progress

Fill in the blanks
(c)

1. Markov models are used to analyze states of _________

system to describe its position at any instant of time.
2. Markov Process is a sequence of experiments in which
each experiment has certain possible ___________.
 Quantitative Techniques for Management Applications

224
Applications

S
Notes
Activity
Discuss the applications of (i) It is a technique applied to solve various management
___________________
markovian technique. problems. For example it can be used to determine the future
___________________ market share of various products/brands. It can be used in
___________________ inventory management also to decide the order size.

E
___________________ (ii) It is very useful for studying the buying pattern of consumers
___________________ or organizations particularly in terms of brand loyalty and
switching patterns.
___________________

UP
___________________ Check Your Progress

___________________
Fill in the blanks:

___________________ 1. It is very useful for studying the _________ pattern of

consumers.
___________________
2. Markov technique is used to solve various ________
problems.

Types
E-
(i) First order Markov process
(ii) Second order Markov process
(iii) Higher order Markov process
Steady State Condition: When state probabilities may become
constant. The system is in steady state condition if following
CC

conditions are satisfied:

(i) The transition matrix elements remain positive from one
period to the next. This property is known as the regular
property of Markov chain.
(ii) It is possible to move from one state to another state in a
finite number of steps, irrespective of the present state.
Absorbing State: A state is said to be absorbing (trapping) state if
it does not leave the state. It occurs when if any transition
probability in the retention diagonal from upper left to lower right
is equal to one (1).
(c)

Transient State: A state is said to be transient if it is not possible

to move to that state from any other state except itself.
Cycling Process: A cycling (periodic) process is one in which
transition matrix contains all zero elements in the retention cells
(diagonal elements) and all other elements are either 0 or 1.
UNIT 17: Markovian Model

Methods for solving Markov Chains 225

S
Notes
Activity
z Transition Probability Matrix
What is transient state?
___________________
z Decision Tree Diagram
___________________
Check Your Progress

E
___________________
Fill in the blanks: ___________________

1. The system is in steady state when state probabilities ___________________

may become ______________.
___________________

UP
2. A _________ (periodic) process is one in which transition ___________________
matrix contains all zero elements in the retention cells.
___________________

Summary ___________________

___________________
Markov models are used to analyze states of stochastic system to
describe its position at any instant of time. It is a technique
applied to solve various management problems. It is very useful for
studying the buying pattern of consumers or organizations
E-
particularly in terms of brand loyalty and switching patterns.
The system is in steady state condition if following conditions are
satisfied:
1. The transition matrix elements remain positive from one
period to the next. This property is known as the regular
property of Markov chain.
CC

2. It is possible to move from one state to another state in a finite

number of steps, irrespective of the present state.

Lesson End Activity

Three brands of toothbrush are available in a provision store. It
has been observed that 50% of customers buy brand Oral-B, 30%
brand Pepsodent and 20% brand Ajanta. The owner of provision
store Sajjan Khatri found that each quarter the customers change
their preference. Of those who bought Oral-B last quarter, 50% buy
it again, but 15% change to brand Pepsodent and 35% to brand
(c)

Ajanta. Of those who bought brand Pepsodent, 70% buy it again,

10% switched to Oral-B and 20% to Ajanta. Of those who bought
brand Ajanta 80% buy it again, 5% switched to Oral-B and 15% to
Pepsodent. Construct the transition matrix and determine their
share in this situation.
 Quantitative Techniques for Management Applications

226
Keywords

S
Notes
Absorbing State: A state is said to be absorbing (trapping) state if
___________________
it does not leave the state.
___________________
Transient State: A state is said to be transient if it is not possible
___________________

E
to move to that state from any other state except itself.
___________________
Cycling Process: A cycling (periodic) process is one in which
___________________ transition matrix contains all zero elements in the retention cells
___________________ (diagonal elements) and all other elements are either 0 or 1.

UP
___________________

___________________
Questions for Discussion
___________________ 1. What are the key properties of Markov models?
___________________ 2. What are the key uses of Markov models?

3. What are the different types of Markov models?

4. What are steady state conditions?

E-
5. Define absorbing state.

6. There are two brands of oil engine A and B. Both have exactly
equal market share in the town presently. The market size is
also fixed. The transition matrix is given below:
To
A B
A [0.8 0.2]
CC

From
B [0.5 0.5]

Determine their future market share for the next year and
market share in the steady state.
7. (a) M/s. Manoj Kumar Kamal Lal stocks three brands of lubes
at its various petrol pumps. Calculate the equilibrium
market share for three brands of lubricants; the transition
matrix is as follows:
To

Castrol Elf Mak

Castrol [0.8 0.1 0.1]
(c)

From Elf [0.05 0.85 0.10]

Mak [0.10 0.06 0.84]

UNIT 17: Markovian Model

227
(b) If the present market share of these three brands is 40%,

S
Notes
30% and 30% respectively, determine their market share
for the year 2008. ___________________

___________________
Further Readings

E
___________________

___________________
Books
___________________
R S Bhardwaj, Mathematics for Economics and Business, Excel
___________________
Books, New Delhi, 2005

UP
___________________
D C Sanchethi and V K Kapoor, Business Mathematics
___________________
Sivayya and Sathya Rao, An Introduction to Business Mathematics
___________________
Web Readings ___________________

www.textbooksonline.tn.nic.in
www.mathbusiness.com
E-
CC
(c)
 Quantitative Techniques for Management Applications

228

S
Notes

___________________

___________________

___________________

E
___________________

___________________

___________________

UP
___________________

___________________

___________________

___________________
E-
CC
(c)
UNIT 18: Data Collection

Unit 18
229

S
Activity
Notes
What is data?

Data Collection
___________________

___________________

E
___________________
Objectives
___________________
After completion of this unit, the students will be aware of the following
topics: ___________________

\ Data characteristics ___________________

UP
\ Requirements of Data in the Organization ___________________
\ Problems and Limitations of Data Collection
___________________
\ Basic Classification of Data
___________________
\ Methods of Data Collection
___________________

Introduction
To solve any managerial problem that you face in the organization
you need relevant information. This relevant information has to
E-
meet the tests of sufficiency and accuracy to be useful to solve the
problem in hand. This information, which is the processed form of
data, refers to collection of numbers, letters, or symbols,
maintained or produced for the management when required.

Data Characteristics
CC

In order, the numbers that you have collected, may be called data,
the following characteristics must be present:
(i) It should be an aggregate of facts; for example, single
unconnected figures cannot be called data, as they cannot be
used to study characteristics of any event or operation of any
industry or organization.
(ii) There should be a reasonable standard of accuracy as is
required for the problem in hand, for example, in the
measurement of length one may measure correctly up to 0.01
cm or 1cm or 1m as required, the quality of the product could
(c)

be reasonably estimated by certain tests of small samples

drawn from big lots of products.
(iii) It should be collected in a systematic manner for a
predetermined objective.
 Quantitative Techniques for Management Applications

230
(iv) The elements of the data must be related to one another. The

S
Notes
Activity
base used in the data should be the same for the data of
What are data characteristics?
___________________
different times or firms to arrive at any meaningful decision.
___________________ For example, you cannot compare two different companies
___________________ figures if one company reports in rupees and another reports

E
in dollars without making the base currency same.
___________________
(v) It must be numerically expressed in measurable units.
___________________
Data, when processed and presented in proper context, becomes
___________________
information which controls the activity of the organization. Data is

UP
___________________
one of the major resources of the organization, developed over a
___________________ period of time and therefore needs to be properly managed and
___________________ safeguarded. It can be treated as inventory because it may be
procured, stored and supplied when needed. Also just like any
___________________
other physical stock it suffers from deterioration and obsolescence.
Data may have different interpretations if not properly defined so
the proper definition is very important. Data also has time
dimension as its use and value will change with time and obsolete
data is not very relevant for information needs. So it is important
E-
to understand the methods of collection of data so that the most
relevant data can be collected and used as soon as possible for the
effective management of the organization.

Check Your Progress

Fill in the blanks:
CC

1. To solve any managerial problem that you face in the

organization you need __________ information.
2. Data, when processed and presented in proper context,
becomes information which __________ the activity of
the organization

Requirements of Data in the Organization

All the managerial processes require information in some form or
the other and therefore, accurate and relevant data is required to
(c)

accomplish almost all the tasks a manager has to perform. The

following tasks are only indicative:
To set the objectives of the company, organization, industry,
Government or any other business entity.
UNIT 18: Data Collection

231
To formulate major strategies and policies to meet specific

S
Notes
objectives.
___________________
To report the result of operations of the business to the share
holders. ___________________

E
___________________
To inform others of selected policies of the company.
___________________
To keep abreast of current operations of the business.
___________________
To inform employees on various matters.
___________________

UP
To prepare long range plans.
___________________
To explore new opportunities.
___________________
To allocate capital resources.
___________________
To exercise necessary control over day-to-day operations.
___________________
To determine the costs underlying various activities of the
firm.
To provide for proper co-ordination and control of business
activities.
E-
All organizations, whether social, political, religious or economic,
are designed to achieve certain objectives. Notwithstanding the
differences in the nature of their activities, the underlying
management processes are common. The management must plan
for and control the usage of various organizational resources
namely, manpower, materials, production facilities and capital in
CC

the most effective manner to achieve the organization’s

objectives. This involves decision making, which is dependent on
the data and its quality. Data management is, therefore, being
increasingly recognized as fifth organizational resource, which
needs to be managed just like other four traditional resources of
man, machine, material and money.

Check Your Progress

Fill in the blanks:
1. Accurate and relevant data is required to accomplish
(c)

almost all the tasks a _________ has to perform.

2. The management must plan for and control the usage of
various ________________ resources.
 Quantitative Techniques for Management Applications

232
Problems and Limitations of Data Collection

S
Notes
The major problems and limitations of data collection can be
___________________
classified under the following broad headings:
___________________
(a) Lack of identification of data needs: The tasks performed
___________________

E
by various levels of management are different. Therefore, data
___________________ needs of the manager vary with the level at which they are
___________________ operating and the function within which they are working.
Very few organizations have made a conscious and deliberate
___________________
effort to identify specific data needs of various managerial

UP
___________________ positions in the organization.
___________________
(b) Response time: This is one of the major data processing
___________________ problems. The data is not collected and processed fast enough
___________________ to allow enough time for mangers to react quickly and in time.
There is a gap between supply of data and requirement of the
data. i.e., Data is supplied with a lead time. In many cases
data is not of any value when it is made available to the
managers.
E-
(c) Inaccessibility of data: Useful and necessary data is
available but is often in a form or location that makes it
uneconomical and infeasible to retrieve.
(d) Differing and conflicting data: Due to different sources
used for collection, data about the some item may differ and
may conflict with each other, for example, two market research
CC

agencies give you a different size of the prospective market for

your products.
(e) Duplication of efforts: Identical data is maintained and
similar reports are generated at several points in the
organization, thereby wasting both time and manpower
resources.
(f) Lack of training: The lack of scientific training in
methodology of data collection is a great handicap in most of
the organizations.
(g) Absence of code of conduct: There does not exist a code of
(c)

conduct for use of data and managers often mould the data in
the way they want to suit their needs without caring for the
accuracy of the same.
UNIT 18: Data Collection

233
(h) Inaccurate and unreliable data: The sheer volume of data

S
Activity
Notes
and human intervention makes it humanly impossible to be
Give an example of primary
consistently accurate and reliable. ___________________
data.
___________________
Check Your Progress

E
___________________
Fill in the blanks:
___________________
1. Due to different sources used for collection, data about
___________________
the some item may differ and may _________with each
other. ___________________

UP
2. There __________ exist a code of conduct for use of data ___________________

and managers often mould the data in the way they ___________________
want to suit their needs without caring for the accuracy
___________________
of the same.
___________________

Basic Classification of Data

Data may be classified as:

(i) Primary
E-
(ii) Secondary

Primary Data

Primary data represents those items that are collected for the first
time and first hand. The data is recorded as observed or
CC

encountered. Essentially, this data is the raw material and may be

combined, or structured in any form. The point to be noted here is
that the data has not been statistically processed. For example,
data obtained by counting the number of bad pieces and good
pieces in the production is the primary unprocessed data. After
this the data can be statistically processed to yield the required
information.

The main advantages of collecting primary data are the following:

(i) They are accurate and reliable as they are collected from the
original source.
(c)

(ii) They provide detailed information according to requirements

of the users.

(iii) It is more reliable and less prone to error.

(iv) Definitions and meaning of terms used in data are explained

to make it understandable and the process transparent.
 Quantitative Techniques for Management Applications

234
(v) Method of collection, its limitations and other aspects are

S
Notes
Activity generally highlighted.
Give suitable example of
___________________
secondary data. Where there are roses there will also be thorns. Following are the
___________________ main limitations of the primary data:
___________________

E
(i) Cost: It is expensive to collect primary data.
___________________
(ii) Time: It is time consuming method of data collection.
___________________

___________________
(iii) Training: It requires experts/trained personnel to collect
data.

UP
___________________

___________________
Secondary Data

___________________ This is also known as published data. Data which is not originally
collected but rather obtained from published sources and is
___________________
normally statistically processed is known as secondary data. For
example, data published by Reserve Bank of India, Ministry of
Economic Affairs, Commerce Ministry as well as international
bodies such as World Bank, Asian Development Bank, etc.
E-
As is the case with primary data there are advantages and
disadvantages associated with secondary data also. The
advantages are:

(i) Cost: It is more economical than primary data, since data is

already available.

(ii) Time: It is faster to collect and process as time has already

CC

been spent to collect the data.

(iii) Information insight: It provides a base on which further

information can be collected to update it and finally use it. It
provides valuable insights and contextual familiarity with the
subject matter.

The limitations of secondary data are as follows:

(i) It may not be too relevant for the problem in hand as it was
originally collected for some other context.

(ii) It could be outdated and hence not of much use in a

(c)

dynamically changing environment.

(iii) The accuracy of secondary data as well as its reliability would

depend on its source as the assumptions made during the data
collection are not specified.
UNIT 18: Data Collection

235
(iv) Locating appropriate source and finally getting access to the

S
Notes
data could be time consuming.
___________________
(v) The data available might be too extensive and a lot of time
___________________
and money may be spent going through it.

E
___________________
Table 18.1: Distinction between Primary Data
and Secondary Data ___________________

Parameter Primary Data Secondary Data ___________________

Source of Data Original source Secondary source
___________________

UP
Method of Data Collection Observation method, Published data from
Questionnaire method, etc. various sources
___________________
Statistical Processing Not Processed Usually processed
___________________
Originality of Data Original Not original. Data collected by
First time collected by user some other agency
___________________
Use of Data Data is compiled for specific There may not be a specific
purpose purpose
___________________
Terms and Definitions of Incorporated May not be incorporated
Units

Copy of the Schedule Included Excluded

Method of Data Collection Given May not be given

Description of Sample Given May not be given
E-
Selection

Time Required More Less

Cost to the Organization Expensive Comparatively cheaper

Efforts Spent More Less

Accuracy of Data More accurate Less accurate

Training Experts/ trained people Less trained personnel

required required

On closer investigation, it will be noticed that the distinction

CC

between primary and secondary data in many cases is of degree

only. Data, which would be secondary in the hands of one, could be
primary for others. For example, to a bank the details of the
customer are primary data, but to a reader of the report of the
bank these details are secondary.

Primary Data
Considerations in Selection of Primary Data Study
While selecting the subject for primary data collection, the
following considerations should be kept in mind:
(c)

(i) Economic Considerations:

(a) Data collection efforts cost money. The value of the

anticipated results must commensurate with the efforts
put in.
 Quantitative Techniques for Management Applications

236
(b) Short-term data collection studies that can yield

S
Notes
appreciable dividends quickly should be preferred to long
___________________ term studies whose benefits may be difficult to foresee.
___________________ (ii) Technical Considerations:
___________________

E
(a) It should be made sure that adequate technical knowledge
___________________ is available to carry out the right process of data collection.
___________________
(b) Where a large problem throws up a number of subjects
___________________ which are independent of each other, it is better to have

UP
___________________ small individual data collected on each subject.

___________________ (c) Where a problem brings to light two or more subjects,

___________________
which are interrelated, independent studies on each might
be carried out in the preliminary stages, but they should
___________________
later be continuously integrated by co-ordinating the
recording of the different teams working on each subject.
The critical examination has to be the completeness of the
data and it has to be carried out by the team as a whole.
E-
(d) The scope and magnitude of the problem would determine
the data required.

(iii) Human Considerations

Where resistance to change or reaction is likely to be there the
data collection should not be proceeded with until acceptance
has been gained.
CC

(iv) Other Limitations and Constraints

(a) Time Limit: Data collection must be completed within time

frame specified so as to be of maximum utilization.

(b) Cost Considerations: Data must be collected within the

cost framework .

(c) Accuracy: Reasonable accuracy, as is required for the

problem, should be ensured.

Methods of Primary Data Collection

(c)

The following four methods of primary data collection are most

widely used:

1. Observation method

2. Personal interview
UNIT 18: Data Collection

3. Questionnaire method 237

S
Activity
Notes
4. Case study method What is observation method?
___________________
Let us look at each method, one by one:
___________________
Observation Method

E
___________________

This is the most commonly used method of data collection, ___________________

especially in studies relating to production management and ___________________
behavioural sciences. Accurate watching and noting down of
___________________
phenomenon, as they occur in nature or at shop floor with regard

UP
to cause and effect, is called the observation method of data ___________________
collection.
___________________
Differentiating characteristics of observation method are as ___________________
follows:
___________________
(i) Direct Method: Direct contacts of sensory organs particularly
eyes and ears are involved to gather and record the data.

(ii) Observe and Record: The observer first observes the

phenomenon carefully and then records data.
E-
(iii) Selective and Purposeful Collection: The observations are
made with a definite purpose in mind and only relevant data
is collected.

(iv) Cause and Effect Relationship: Observation method leads

to development cause and effect relationship.
CC

Box 18.1: Merits and limitations of Observation Method

Merits

This method of observation is common to all the discipline of

research is simple to use.
It is realistic as it is based on actual and first hand
experience.
The conclusions are more accurate reliable and dependable.
This method is used for formulation of hypothesis.
This method is successfully used for verification of hypothesis.
It is useful when in-depth study is required.
(c)

Limitations
Some events cannot be observed without biases. For example,
it is not possible to observe emotions and sentimental factors,
like and dislikes without bias about the degree of emotions.
Contd….
 Quantitative Techniques for Management Applications

238 It sometimes results in illusory observations.

S
Notes
Being a long drawn process, the techniques of observation are
___________________ expensive and time consuming.
___________________ Sometimes the atmosphere tends to become artificial and this
leads to a sense of self-consciousness among the individuals
___________________

E
who are being observed. This defeats the purpose of
___________________ observation.
___________________ The slowness of observation methods leads to disheartening
___________________ and disinterest among both the observer and observed.

UP
___________________ The final results of observation depend upon the
interpretation and understanding of the observer, the defects
___________________
of the subjectivity in the explanation creep in the description
___________________ of the observed and deductions from it.

___________________ As the purpose of the observation is known to observers,

therefore, it is their own wish to record or view a particular
thing.

The control can be of two forms. The observer could be a

E-
participant or a silent observer. In group discussions he is
normally a silent observer but in interview techniques he becomes
a part of the interview and hence his lack of objectivity may
hamper the quality of his observations.

Controlled and uncontrolled observation methods are the two sub-

methods used to watch and understand the observation.
CC

(i) Controlled Observation: This is a systematic observation

based on logic and reasoning. This is done on a preconceived
plan and deliberate effort is made to control the phenomenon.

(ii) Uncontrolled Observation: In this method observations are

made in a natural surrounding. There is no planning, no
control and no use of any deliberate effort to change the
working of the phenomenon.

Table 18.2: Distinction between Controlled and

Uncontrolled Observation
Parameter Controlled Uncontrolled
Observation Observation
(c)

Control Dimensions Control over the pheno- No control. Observations

menon, conditions of light, under natural conditions
temperature, humidity, etc.
Control over the observer or
observed
Techniques of Control Planning of observations No need to use control
Used situations, Use of mechanical techniques
appliances such as recorders,
Contd….
UNIT 18: Data Collection

watch, etc. 239

Maps and sociometric scales

S
Hypothesis Activity
Notes
Detailed notes Discuss the main points in
Group discussions ___________________
personnel interview.
Degree of Bias Subjective study and bias This is an objective study
comes in during study and keeps the
___________________
observations bias free

E
___________________
Cause and Effect Well established Difficult to establish
Relationship ___________________
Degree of Reliability High Low
of Data ___________________

___________________
The process of observation method is used most effectively in the

UP
field observations where the presence of the observer does not ___________________
make a difference to the observed. For example, if you want to ___________________
know how many people enter the New Delhi railway station from
___________________
the Paharganj side, you just have to stand at the gate and count.
Your presence there or not being there does not matter to people ___________________
who are being observed.

Steps in Organization of Field Observation

Following are the main steps generally followed in the

E-
organization of field observations:

(i) Determination of nature and limits of observation:

Depending upon the nature of research and hypothesis, an
outline of the research is prepared. This helps the observer to
guide him on what should be observed and on what should be
left out.
CC

(ii) Determination of time, place and subject of study: A

project can be of short or long duration, it may be studied
under laboratory conditions or in the open. It should also be
decided that whether we shall observe the behaviour of
phenomenon as a whole or of the individual items in relation
to the total.

(iii) Determination of the investigators: Depending on the

nature, work content and objectives, the needs for individual
investigator or of a team are to be identified.

(iv) Provision of mechanical appliances needed: The

(c)

mechanical appliances for recording such as tape recorder,

movie camera, etc., required should be identified in the
beginning and used when needed.

When you take care of these basic steps, your data would be useful
and relevant to the problem in hand.
 Quantitative Techniques for Management Applications

240
Personal Interview Method

S
Notes
Under this method of collecting data there is a face to face contact
___________________ with the persons from whom the information is to be obtained
___________________ (known as informants). The interviewer asks them questions
pertaining to the survey and collects the desired information. For
___________________

E
example, if a person wants to collect data about the working
___________________
conditions of the workers of Hindustan Lever Ltd., Mumbai, he
___________________ would go to the HLL factory at Mumbai, contact the workers and
___________________ obtain the required information. The information obtained is direct

UP
and original. This is the most suitable method of data collection for
___________________
business and economic problems.
___________________
Table 18.3: Merits and Demerits of Personal Interview Method
___________________
Merits Demerits
___________________
In this method, direct contact between There are certain matters, which
researcher and informants is established can be written in privacy but
and effective communication is built, about which one does not wish to
which helps in getting direct information speak before others. If these
about paradigms, inner feelings, matters are the subjects of
E-
emotions and sentiments. interview, the likelihood is that
only a disguised version of these
will be presented.

Fine tuning of the responses can be done If an interviewee is of low level

so as to get out the best possible by intelligence he is usually unfit to
rephrasing the questions and probing give correct information. Same
deeper wherever required goes for interviewer also as
interviewing is an art rather than
CC

science and the art has to be

mastered

An interview gives us knowledge of facts, If the interviewer is unable to

which are inaccessible to observation. suppress his prejudices, his
The emotional attitude, secret motivation understanding and interpretation
and incentives governing human life of data given in the interview will
come to surface in an interview though be defective
these are unobservable. Therefore,
interview has a quality which may be
called supra-observational

Through this method it is possible to In the interview, certain aspects

verify the information that has been of the human behaviour get
(c)

collected from other sources. overemphasized at the expense of

others. There is a tendency to give
too much importance to personal
factors and minimize the role of
environmental factors. This has to
be guarded against.
UNIT 18: Data Collection

Types of Interviews 241

S
Notes
Activity
Interviews can be classified according to their basic characteristics.
What are the different types of
___________________
(i) According to Formalness: interview?
___________________
(a) Formal Interview: In formal interviews, the interviewer

E
___________________
presents a set of well defined questions and notes down
the answers of informants in accordance with the ___________________
prescribed rules. Here, emphasis is given on the order and ___________________
on sequence of question.
___________________

UP
(b) Informal Interview: Here the interviewer has the freedom
___________________
of alterations in questions to suit a particular situation in
formal interview. He may revise, reorder or rephrase the ___________________
questions to suit the needs of the respondents. The ___________________
emphasis is on situation and questioning generally
depends on the situation and individual. ___________________

(ii) According to Number:

(a) Personal Interview: In personal interview only a single

person is interviewed at one time. Detailed knowledge
E-
about intimate and personal aspects of individual can be
obtained as it is face-to-face talk.

(b) Group Interview: In this method two or more persons are

interviewed at the same time. The group interview is,
therefore, more suited for gathering routine information
rather than personal information.
CC

(iii) According to Purpose:

(a) Diagnostic Interview: In this type of interview, interviewer

tries to understand the cause or causes because of which a
particular fact or incident happened. For example,
diagnostic interviews are held with the operators with a
purpose to grasp the cause and nature of failure of
machines and not to ascertain whether failure has
occurred.

(b) Research Interview: These interviews are held to gather

information pertaining to certain problems but may not be
(c)

as specific as diagnostic interviews. The questions to be

asked to gather the desired information are
predetermined. In as much as this data is gathered for the
purpose of research into a problem, this is called research
interview.
 Quantitative Techniques for Management Applications

242
(iv) On the basis of Function and Methodology:

S
Notes
(a) Non–directed Interview (Non Directional Interview): This is
___________________
also known as free or unstructured interview. This is a
___________________ type of interview in which the interviewer exercises no
___________________ control, provides no direction and has no brief or

E
predetermined set of questions to ask. The interviewer
___________________
merely engages the interviewee in talks and encourages
___________________ him to tell about his experiences and feelings. This type of
___________________
interview is suitable when the researcher wishes to assess
the amount of awareness a person has about certain

UP
___________________
problems and the manner in which he views them.
___________________
(b) Focussed Interview: This method is employed for studying
___________________ the socio-psychological effects of mass media like radio,
___________________ television, cinema, etc. The specialty of the focused
interview is that by its means the personal reactions,
emotions and intellectual orientation of the persons to be
interviewed towards specific issues can be studied.

(v) Classification according to Subject Matter:

E-
(a) Qualitative Interview: The qualitative interviews are about
complex and non-quantifiable subject matter. For example,
interviews held for case studies for specific problem study
are qualitative, because the interviewer has to cover past,
present and future to know a case. In this a qualitative
analysis associated with a situation is performed. The
CC

subjective opinion of the interviewer is sought.

(b) Quantitative Interview: The quantitative interviews are

those in which certain set facts are gathered about large
number of cases. The census interviews are an example of
this type.

Many combinations of these types can be made to suit a particular

situation.

Getting Correct Response in an Interview

The main concern of the researcher employing the method of

(c)

interview is to get correct and to the point answers to the topic of

research. A research can be less expensive and economical only if
deviations from the main line of approach are kept under control.
Normally, the accuracy of the responses depends upon the skill and
tactful approach of the interviewer and no rules can be framed in
UNIT 18: Data Collection

this connection. Still the following points can be kept at the back of 243

S
your mind: Notes

Prior to start the business with the interviewee, interviewer ___________________

must develop rapport with the interviewer, so that he feels ___________________
comfortable with him.

E
___________________
For allowing maximum opportunity of self-expression to the
___________________
interviewee, he should be allowed to narrate his experience in
the story form. ___________________

The interviewee and interviewer should be free and frank. The ___________________

UP
interviewee should be allowed to describe whatever he thinks ___________________
worthwhile. Even if some irrelevant facts are being described
___________________
the interviewee need not be checked. He should not be
discouraged. Though maximum freedom of self-expression is ___________________
desirable, this can only be within the scope of the problem ___________________
being discussed. This requires alertness and direction at the
suitable occasion. Good humour is the essence of successful
direction.
The interviewer must hear the interviewee with full interest.
E-
Nobody should be able to guess from his expression that he is
bored or his mind is elsewhere.
If an interviewer can convince the interviewee that he
appreciates his cooperation and greatly values the information
given by him, this word of encouragement has a salutary
effect on the interviewee, who then gives more focussed
CC

responses.
The information given by the interviewee, if suspected, can be
tested through cross-examination of the interviewee.
Moreover, the emotional expression accompanying the
responses give a clue to the interviewer about the veracity or
otherwise of the answer being given.

Precautions to be taken while Interviewing

Following are the main causes, which render an interview

unusable. These should be taken care of when interviewing:
(c)

(i) Often interviewees, under emotional spells, exaggerate the

facts in order to satisfy their vanities and create impression. It
should be taken with a pinch of salt.

(ii) Sometimes there is communication gap between the

interviewer and interviewee with the result that interviewees
 Quantitative Techniques for Management Applications

244
say one thing and the interviewer understands

S
Notes
something else.
___________________
(iii) Some interviewees deliberately try to mislead the interviewer
___________________
and make a fool of him, the interviewer must be mature and
___________________ experienced enough to tell off and rebuff such fake

E
___________________ interviewees.

___________________ (iv) Sometimes an inexperienced interviewer is offended by the

___________________ behaviour of interviewees and in a revengeful mood distorts

UP
the facts in his report.
___________________

___________________ (v) Interviewer should critically examine those aspects of the

interview in which the relationship of cause and effect seems
___________________
to hold. This helps to determine whether the causes are
___________________ always present or not when certain effects appear.

Questionnaire Method
Under this method, a formal list of questions pertaining to the
survey (known as questionnaire) is prepared and sent to the
E-
various informants. Questionnaire contains the questions and
provides the space for answers. A request is made to the
respondents through a covering letter to fill up the questionnaire
and send it back within a specified time.

The questionnaires could be structured or unstructured.

Structured questionnaires are those that pose definite, concrete
CC

and preordained questions with fixed response categories. In

unstructured questionnaires, questions are not necessarily
presented to the respondents in the same wording and do not have
fixed responses. Respondents are free to answer the questions the
way they like in their own wording and style. Questionnaires could
be a mix of the two types also leaving the field wide open to the
designer of the questionnaire.

Types of Questions used in Devising a Questionnaire

Dichotomous Questions: When reply to a question is in the

form of one out of two alternatives given, one answer being
given in negative and other positive, it is called a dichotomous
(c)

question. Both the negative and positive answers combined

together form the whole range of answers given. For example:
“Whether respondent is educated………………..Yes/No.”
Multiple Choice Questions: In these questions normally
three to five alternative answers are given. These alternatives
UNIT 18: Data Collection

are quite comprehensive and the respondent has to select one 245

S
of them. In framing these types of questions, the framer has to Notes
Activity
be cautious enough that all the possible alternatives are What are Dichotomous
___________________
included in it and they are mutually exclusive. Questions?
___________________
Ranking Item Questions: A variation on multiple choice

E
___________________
questions, these questions are so designed as to record the
preferences of the respondent. In ranking item questions there ___________________
may be several preferences arranged item wise. ___________________
Open-ended Questions: Questions, which are of descriptive ___________________

UP
type and allow the respondent to cite his experiences, are
___________________
known as open-ended questions.
___________________
Leading Questions: These are suggestive questions. In these
types of questions the reply is suggested in a particular ___________________
direction. They should be avoided as far as possible. ___________________
Ambiguous Questions: The questions that lack clarity and
are so worded that the meaning is not clear are known as
ambiguous questions. Such questions normally should not be
included in the questionnaire as they are likely to confuse the
E-
respondent. The meaning of such questions is not uniformly
convulsed to all the respondents.
Table 18.4: Merits and Demerits of Questionnaire Method

Merits Demerits
In comparison to other methods, Lack of interest on the part of
the questionnaire method is both respondents lowers the number of
cheaper and quicker responses, making the study
CC

unreliable
It requires less skill to administer Incomplete and illegible
than other methods responses renders the whole
response bad
If the informants or the If a problem requires deep and
respondents are scattered in large long study, it cannot be studied
geographical areas, this is the through this method
most suitable method
Besides saving money, time is This method is very rigid since no
also saved as simultaneously alteration and rephrasing of
hundreds of persons can be questions can be used
approached
It is more reliable in special cases Prejudices and bias of the
although in most cases the researcher influences the framing
(c)

reliability is suspected of the questions

The respondent is free from Sometimes the questionnaire is
external influences, such as itself incomplete and leaves out
researcher and therefore provides certain critical questions which
reliable, valid and meaningful are unearthed later rendering the
information whole exercise fruitless

Contd…
 Quantitative Techniques for Management Applications

246 Chances of errors are low because There is no provision in this

S
Notes respondent supplies information method for coming face to face
himself with the respondent. This may
___________________ result in manipulation of replies
by the respondents.
___________________
The informants are directly
___________________ involved in the supply of

E
information, so the method is
___________________ more original
The impersonal nature of
___________________ questionnaire ensures uniformity
from one measurement situation
___________________
to another.

UP
___________________
Considerations in Questionnaire Design
___________________

___________________ Questionnaire is always framed with the help of certain

background material and the problem statement. The first
___________________
requirement always is the design of the problem statement and
this is the area where most of the questionnaires go wrong. If your
problem statement is faulty, your questions are not going to point
to the required direction and you are bound to get wrong
inferences. One should spend the maximum time on it, since it will
E-
be well spent.

After the problem statement comes, the issue of the respondents as

their intellectual level has to be kept in mind while designing
questions. If the questions, language and wordings are not in
accordance with the intellectual level of respondents then it would
not be possible for them to furnish correct replies. In such a
CC

situation the purpose of the research would not be fulfilled. The

outcome of past experiences enables the researcher to know the
shortcomings beforehand, enabling him to remove these
deficiencies so as to improve the response rate.

Other factors to be taken into account in the construction of a

questionnaire:

Appeal: Each questionnaire should be attached with an

appeal in which the aim and purpose of the questionnaire is
set forth and the sincere co-operation of the respondents is
requested. The appeal may be made more effective by giving
appropriate incentives in the form of money, books, and with a
(c)

promise to give a copy of the report to the respondents.

Instructions for filling up the questionnaire: The

questionnaire must carry a list of instructions for filling it up
and dispatching it. The respondent must not have to pay for
return postage, unless you are promising a prize for
UNIT 18: Data Collection

responses. If the questionnaire is time bound, the last date of 247

S
receiving completed response and the address should be Notes
Activity
clearly written. What are the problems of
___________________
response?
Clarity of questions: For desired response it is of utmost ___________________
importance to formulate questions that are direct, clear and

E
___________________
precise.
___________________
Order of questions: The questions should be broken up into
___________________
sections and each section should have a number of questions,
which are mutually interrelated. Question about personal ___________________

UP
detail should be avoided or should be asked in the end. ___________________

Protesting of Response in Questionnaire ___________________

The basic thing that has to be kept in mind is that ambiguity ___________________

should be avoided in collecting data through questionnaire method. ___________________

For this, it is necessary that the questionnaire should be tested
before it is actually used in a business research study. Pre-testing
is nothing but testing of questionnaire before it is actually used. If
testing is to be done the right way the following steps are required:
E-
Testing the validity in a representative sample: The
questionnaire should be tested in every respect, before it is
actually mailed to the target segment. This testing can be
done on a limited number of people through sampling method
but while testing it within the sample, it should not be
forgotten that the sample should be perfectly representative of
CC

the target segment.

Protesting to check whether the results are in tune with

objectives: The questionnaire should meet the objective of
research study. It means that it should help in getting
maximum possible relevant responses. It is, therefore,
necessary that it should be made suitable to objectives of
study even if it requires testing more than once.

Poor response requires modification of the

questionnaire: The questionnaire is mailed to the informants
who are required to fill it and send it back. If the response of
(c)

the informant is poor and very few questionnaires are

returned, it means that there is something wrong with the
form and style of the questionnaire and it requires
modifications/change and reframing. Furthermore, if the
questionnaires returned are incomplete or the replies are not
 Quantitative Techniques for Management Applications

248 satisfactory and up to the mark, it should be presumed that

S
Notes the questionnaire is defective and it requires modification.
___________________ After modification the questionnaire should again be
subjected to pre-testing.
___________________

___________________ Problem of Response: Difficult Situations for the Researcher

E
___________________
When the questionnaire is not leading to any response, one of the
___________________ following factors is usually responsible for it:
___________________
Importance of the problem to the respondents: It is

UP
___________________
generally seen that those who are concerned with the problem
___________________ give better response than those who are not.
___________________
Characteristics of the respondents and prestige of the
___________________ sponsoring body: It is seen that educated people with social
consciousness are more responsive as compared to people
belonging to lower economic group. If the research study is
sponsored by a well-known organization it is likely to have
better response.
E-
Form and nature of questionnaire and arrangements of
the questions: Questionnaire also plays its part in the matter
of response. If the questionnaire is short and has been printed
in attractive manner, its layout is neat and attractive, the
arrangements of questions are scientifically planned, and it is
CC

likely to invite a better response.

To get better response, inducement is needed. Inducement may be

classified under two heads: monetary and non-monetary. Monetary
inducement is given generally to people who are economically weak
or likely to be influenced by money. This money is given in advance
or after receiving the filled questionnaire. Non-monetary
inducement may be in the form of a reward. It may be a letter of
appreciation or mentioning of the name in the report of study and
so on. The suitability of the inducement to the study and the
respondent’s expectations has to be kept in mind when deciding
(c)

upon which inducement to use.

To take care of the poor response situations companies normally

get students to help to find respondents and to filling up the
questionnaires. Excellent way for you to make money while
studying!
UNIT 18: Data Collection

Schedule 249

S
Notes
Schedule is a variation of the questionnaire and can be defined as
___________________
a proforma that contains a set of questions which are asked and
filled by an interviewer himself in a face-to-face situation with ___________________
interviewee. Unlike a questionnaire, the schedule acts as a

E
___________________
guideline to the interviewer trying to get the required response
___________________
from the interviewee. Schedule is a standardized device or a tool of
observation to collect data in an objective manner. Same guidelines ___________________

as mentioned in the questionnaire are to be kept in mind while ___________________

UP
making these schedules.
___________________

Case Study Method ___________________

Case study method may be defined as small, inclusive and ___________________

intensive study of a situation in which investigator uses all his ___________________
skills and methods for systematic gathering of enough information
about a situation to understand the problem and its solution. The
case study is a form of qualitative cum quantitative analysis
involving the very careful and complete observation of a person,
E-
situation or institution.

Table 18.5: Merits and Demerits of Case Study Method

Merits Demerits
Intensive and deep study of Several unrealistic assumptions may
the problem is possible be made when structuring the case,
making it difficult to relax them later
on
CC

Study of subjective aspects of It is expensive in terms of money,

the problem is possible and time and energy
more elaborative than other
methods
Comparison of possible If there is improper understanding
problem statements is easier between the developer and the
respondents, the data and hence the
inferences could be false and
misleading
Valid hypothesis can be Prejudices and biases come in more
formulated and tested while easily as the study is more subjective
the case is in development
Is very useful when you have It is not possible to apply sampling
to study processes and not methods and generalization often
(c)

isolated incidents leads to false conclusions.

Very useful in situations
where more of qualitative
rather than quantitative
decision making is involved.
 Quantitative Techniques for Management Applications

250
Check Your Progress

S
Notes
Fill in the blanks:
___________________
1. _____________ is a variation of the questionnaire and
___________________
can be defined as a proforma that contains a set of
___________________ questions which are asked and filled by an interviewer

E
___________________ himself in a face-to-face situation with interviewee.
2. _____________method may be defined as small, inclusive
___________________
and intensive study of a situation in which investigator
___________________ uses all his skills and methods for systematic gathering

UP
___________________ of enough information about a situation to understand
the problem and its solution.
___________________

___________________
Secondary Data
___________________
Although primary data is required for most of the internal
business situations, many of the strategic decisions depend upon
the information that is external to the organization. The criticality
of the decision and the time factor involved would decide whether
E-
secondary data is to be used or the situation calls for primary data.

If the situation calls for secondary data, this data would normally
be either published or unpublished. Unpublished records, although
dealing with the matters of public interest, are not available to
people in published form. It means that everybody cannot have
CC

access to these records. Proceedings of the meetings, noting on the

files, private research, etc., form the category of unpublished
records. Normally these records are very reliable since there is no
fear of their being made public; the writers give out their views
clearly.

Published records are available to people for investigation, perusal

and for further use, survey reports, magazine articles, published
studies, etc., fall under this category. The data contained in these
documents can be considered reliable or unreliable depending upon
the agency that is collecting the data and the sources it had used
(c)

for collecting this data. Most of the information that is now

available to people and researchers in regard to business
environment are to be found in the form of reports. The reports
published by governments are considered more dependable on one
hand and on the other hand some people think that the reports
UNIT 18: Data Collection

251
that are published by certain private individuals and agencies are

S
Notes
more dependable and reliable.
___________________
There are so many sources of published data that it is impossible
___________________
to name them all here. In spite of so many sources, the published

E
data usually suffers from the following drawbacks: ___________________

___________________
(a) Data about all the aspects of business and economic activity
___________________
are not collected.
___________________

UP
(b) Even the Government of India does not have an up-to-date
___________________
and latest data about many socio-economic aspects as well as
the business environment, although it is now working towards ___________________

it. ___________________

___________________
(c) Data lacks in homogeneity and continuity.

(d) The data collected by the Government agencies is not beyond

doubt. This is due to the approach of the administration and
also because of the method of data collection. The resources
E-
that are put at the disposal of the machinery that is entrusted
for the task of collection of data are very meagre.

(e) Data collected by private agencies run the risk of their biases
coming into picture, as also their own aims and objectives
could make them present the data in an improper way
CC

rendering it unuseful for you.

Therefore, before using the secondary data, it is essential that the

investigator should satisfy himself that the data is: (a) Reliable, (b)
Suitable, (c) Adequate and (d) Timely. Reliability of data can be
established by asking yourself the following questions: Who
collected the data and from which sources? Are the methods used
in collecting are standard methods and reliable? Whether both the
compiler and source are dependable? The purpose for which the
data were originally collected is in tune with the purpose that you
are going to use the data for; the secondary data should be suitable
(c)

for the purpose of enquiry. Even if the data is reliable it should not
be used if the same is found to be unsuitable for the enquiry. For
checking the suitability of data one should see: What was the
object of the enquiry? The definitions of various items and units of
collection must be carefully scrutinized. What was the accuracy
 Quantitative Techniques for Management Applications

252
aimed at? What is the time of collection of data required? Can it be

S
Notes
regarded a normal time? Is the data homogeneous?
___________________
The secondary data may be reliable and suitable but the same may
___________________
be inadequate for the purpose of investigation. The data collected
___________________

E
earlier may refer to a problem area which could be narrower or
___________________ wider than the area required for the present enquiry and if it is
___________________ such, the data should be carefully scrutinized to test whether it
___________________ meets the requirements or not. If it does not meet the

UP
requirements of the scope or the time frame of study, do not use
___________________
the data just because it is the only data that is there. Although
___________________
knowledge of the matter under consideration and proper use of the
___________________ statistical methods is presupposed, great care is necessary in
___________________ dealing with published statistics because of the limitations or
inaccuracies that may be present.

Check Your Progress

Fill in the blanks:
E-
1. If the situation calls for _____________ data, this data
would normally be either published or unpublished.
2. ___________ records, although dealing with the matters
of public interest, are not available to people in
published form.
CC

3. The cost of collection of ................... is much higher than

the collection of secondary data.
4. Under ................... method, the investigator collects data
from a third party or witness or head of an institution,
etc.
5. The collected data are arranged into ................... groups.
6. A classification is said to be ................... if it brings out
essential features of the collected data.

Summary
(c)

Any set of numerical figures cannot be regarded as statistics or

data. A set of numerical figures collected for the investigation of a
given problem can be regarded as data only if these are comparable
and affected by a multiplicity of factors. Data from a primary
source are original and correspond to the objective of investigation.
UNIT 18: Data Collection

However, the secondary data are often available in published form, 253

S
collected originally by some other agency with a similar or Notes
different objective. Generally, the primary data are more reliable ___________________
than the secondary data which, however, are more economical.
___________________

E
___________________
Lesson End Activity
___________________
How will you design a questionnaire?
___________________

Keywords ___________________

UP
___________________
Data: The facts and figures that is collected, analyzed and
___________________
interpreted.
___________________
Observation: The set of measurements obtained for a single
element in the data set. ___________________

Qualitative Data: Data that are labels or names used to identify

an attribute of each element. Qualitative data use the nominal or
ordinal scale of measurement and may be non-numeric or numeric.
E-
Statistics: The art and science of collecting, analyzing, presenting
and interpreting data.
Quantitative Data: Data that indicate that how much or how
many of an element. Quantitative data use the interval or ratio
scale of measurement and are always numeric.
CC

Questions for Discussion

1. Define data. What are the key characteristics of data?

2. Why data is required in the organisations?

3. What are the key problems in data collections?

4. What are the different types of data?

5. Make a distinction between primary and secondary data.

6. What are the key methods of collecting primary data?

7. What are the different types of interviews?

(c)

8. What are the key drawbacks of collecting data from secondary

sources?
 Quantitative Techniques for Management Applications

254
Further Readings

S
Notes

___________________ Books
___________________ R S Bhardwaj, Mathematics for Economics and Business, Excel
___________________ Books, New Delhi, 2005

E
___________________ D C Sanchethi and V K Kapoor, Business Mathematics

___________________ Sivayya and Sathya Rao, An Introduction to Business Mathematics

___________________
Web Readings

UP
___________________
http://www.math.ucla.edu
___________________
www.textbooksonline.tn.nic.in
___________________

___________________
www.mathbusiness.com
E-
CC
(c)
UNIT 19: Presentation of Data

Unit 19
255

S
Activity
Notes
What are numeric data?

Presentation of Data
___________________

___________________

E
___________________
Objectives
___________________
After completion of this unit, the students will be aware of the following
topics: ___________________

\ Categorical data and numeric data ___________________

UP
\ Utility and Advantages of Diagrammatic Presentation ___________________
\ Numeric data
___________________
\ Categorical data
___________________

___________________
Introduction
An important function of statistics is the presentation of complex
mass of data in a simple way so that it becomes easier to
understand. Classification and tabulation are the techniques that
E-
help in presenting the data in an intelligible form. But with
increase in volume of data, it becomes more and more inconvenient
to understand even after its classification and tabulation. Thus, to
understand various trends of the data at a glance and to facilitate
the comparison of various situations, the data are presented in the
form of diagrams and graphs.
CC

Categorical Data and Numeric Data

Both categorical data and numeric data can be used to create
statistics.
1. Numerical Data: Numeric data are data that exist in
numeric form, such as height, the number of children in a
household and annual income. Numeric data is most definitely
quantitative, and can be either "continuous," such as time, or
income, or "discrete," such as number of children.
2. Categorical Data: Categorical data are data that take a
(c)

finite set of values that can be either numeric or categorical.

For example, eye colour, ethnicity, and country of residence
are all possible types of categorical data that are not numeric.
 Quantitative Techniques for Management Applications

256
Statisticians often refer to categorical data as "qualitative data,"

S
Notes
while social scientists might consider categorical data to be
___________________
"quantitative."
___________________
Check Your Progress
___________________

E
Fill in the blanks:
___________________
1. Numeric data are data that exist in ______ form, such as
___________________
height, the number of children in a household and
___________________
annual income.

UP
___________________
2. Statisticians often refer to categorical data as
___________________ ___________ data
___________________
Utility and Advantages of Diagrammatic Presentation
___________________
Data presented in the form of diagrams are useful as well as
advantageous in many ways, as is obvious from the following:
1. Diagrams are attractive and impressive: Data presented
in the form of diagrams are able to attract the attention of
E-
even a common man. It may be difficult for a common man to
understand and remember the data presented in the form of
figures but diagrams create a lasting impression upon his
mind. Due to their attractive and impressive character, the
diagrams are very frequently used by various newspapers and
magazines for the explanation of certain phenomena.
Diagrams are also useful in modern advertising campaign.
CC

2. Diagrams simplify data: Diagrams are used to represent a

huge mass of complex data in simplified and intelligible form
which is easy to understand.
3. Diagrams give more information: In addition to the
depiction of the characteristics of data, the diagrams may
bring out other hidden facts and relations which are not
possible to know from the classified and tabulated data.
4. Diagrams save time and labour: A lot of time is required to
study the trend and significance of voluminous data. The same
data, when presented in the form of diagrams, can be
understood in practically no time.
(c)

5. Diagrams are useful in making comparisons: Many a

times the objective of the investigation is to compare two or
more situations either with respect to time or places. The task
of comparison can be very conventionally done by the use of
diagrams.
UNIT 19: Presentation of Data

257
6. Diagrams have universal applicability: Diagrams are

S
used in almost in every field of study like economics, business, Notes

administration, social institutions and other fields. ___________________

___________________
Check Your Progress

E
___________________
Fill in the blanks:
___________________
1. Data presented in the form of diagrams are able to
attract the ___________of even a common man. ___________________

___________________
2. Diagrams are used in almost in _____________ of study

UP
like economics, business, administration, social ___________________
institutions and other fields. ___________________

3. _____________ is a systematic presentation of numerical ___________________

data in rows and columns.
___________________

Summary
Presentation of data is provided through tables and charts. A
E-
frequency distribution is the principal tabular summary of either
discrete or continuous data.

Lesson End Activity

Collect data on IOC, HPCL, BPCL & ONGC company’s financial
performance from different sources like internet, newspapers,
CC

magazines, etc. Are there any differences between the same? What
inferences do you draw on the objectives of that particular type of
media when they are presenting data?

Keywords
Data Presentation: The presentation of complex mass of data in
a simple way so that it becomes easier to understand.
Frequency Distribution: A frequency distribution is the principal
tabular summary of either discrete or continuous data.
Numerical Data: Numeric data are data that exist in numeric
(c)

form, such as height, the number of children in a household and

annual income.
Categorical Data: Categorical data are data that take a finite set
of values that can be either numeric or categorical.
 Quantitative Techniques for Management Applications

258
Questions for Discussion

S
Notes
1. What is data presentation means?
___________________

___________________
2. What is frequency distribution meant for?

___________________ 3. What is numerical data?

E
___________________ 4. What is Categorical data?

___________________ 5. What are the different techniques of data presentation?

___________________ 6. Try and collect as much data as possible from different sources

UP
___________________
about the health levels of the people residing in your area.
What problems come in while collecting this data?
___________________
7. If you have to use the sampling method in question 6 what
___________________ method would you use and how would you reduce sampling
___________________ and non-sampling errors in your sample?
8. Ansal Builders is engaged in the construction of a multi-storey
building for setting up a lube factory. It has recently conducted
a cost audit. The manager (cost accounting) has collected the
figures of total cost and its major constituents. The
E-
information collected as percentage of expenditure is shown
below. Represent the data with the help of a suitable diagram.
Item Expenditure %
Wages 25
Bricks 15
Cement 20
CC

Steel 15
Wood 10
Supervision and Misc. 15

9. Chand Contractors supplies contract labour to various

industrial units for carrying out their various production
activities in and around Bhilai. Mr. R.B. Tripathi is the chief
consultant and is responsible to manage the continuous
supply of contract labour on weekly basis. The daily wages of
contract labour varies from ` 25 to 95 per day depending on
the skill, experience and the nature of work in the industry
utilizing the services of contract labourers. The daily wages
and number of workers data have been compiled by Shri
(c)

Tripathi for estimating the number of workers demanded and

their average wages.
Draw a suitable diagram of the data to enable the chief
executive of Chand Contractors to understand the relations
between wages and number of workers.
UNIT 19: Presentation of Data

259
Find out the number of workers getting wages lower than 57

S
Notes
and more than 77 using Ogive graphs.
___________________
Daily No. of Daily No. of
Wages (`) Workers Wages (`) workers ___________________
20-25 21 60-65 36

E
___________________
25-30 29 65-70 45
30-35 19 70-75 27 ___________________

35-40 39 75-80 48 ___________________

40-45 43 80-85 21
___________________

UP
45-50 94 85-90 12
50-55 73 90-95 5 ___________________

55-60 68 ___________________

10. IBM Computers (I) Ltd. has been entrusted with the ___________________
responsibility of developing a relationship between number of ___________________
employees and salary structure in Arian Pharmaceuticals Ltd.
The statistics manager, Mr. Ayyar has collected the following
data. Draw the frequency distribution and superimpose
frequency polygon and frequency curve on it.
E-
Salary No. of Employees Salary No. of Employees
300-400 20 700-800 115
400-500 30 800-900 100
500-600 60 900-1000 60
600-700 75 1000-1200 40

Further Readings
CC

Books
R S Bhardwaj, Mathematics for Economics and Business, Excel
Books, New Delhi, 2005

D C Sanchethi and V K Kapoor, Business Mathematics

Sivayya and Sathya Rao, An Introduction to Business Mathematics

Web Readings
http://www.math.ucla.edu
(c)

www.textbooksonline.tn.nic.in

www.mathbusiness.com
 Quantitative Techniques for Management Applications

260

S
Notes

___________________

___________________

___________________

E
___________________

___________________

___________________

UP
___________________

___________________

___________________

___________________
E-
CC
(c)
UNIT 20: Case Studies

Unit 20
261

S
Notes

Case Studies
___________________

___________________

E
___________________
Objectives
___________________
After analyzing these cases, the student will have an appreciation of the
concept of topics studied in this Block. ___________________

___________________

UP
___________________
Case Study 1: ASSOCHAM
___________________
Associated Chamber of Commerce and Industry (ASSOCHAM) is
very much concerned about the employment of youths and their ___________________
pay rolls in small oil an industries, with special reference to
___________________
ancillary parts manufacturing, transport for hire, taxis, dealers of
new and old vehicles, petrol stations and automobile repair
garages. The chamber has employed you to collect the data
regarding employment and payroll as on 31st March, 2000 and
present it suitably through diagram so that it can be include in
E-
the final memorandum to be submitted to the Minister for
Industries.

The data that you have collected is as follows:

Avg. Earnings
Employment
Industry per employee
on 31-3-2000
per year (`)
CC

1. Parts manufacturers 4,34,856 56,540

2. Transport for hire 15,26,897 26,348

3. Taxis 11,32,560 42,685

4. Dealers of new and 1,09,805 13,684
used vehicles
5. Retail filling stations 22,25,960 15,008
6. Automobile repair 12,35,200 12,048
garages

Question

Present the data using a suitable diagram(s) so as to bring out the

(c)

finer points.
Source: MBCQ-721D_Quantitative Techniques for Management Applications_CCE_UPES
 Quantitative Techniques for Management Applications

262
Case Study 2: Mount Shivalik

S
Notes
Mount Shivalik Distilleries is a progressive manufacturer of
___________________ ‘Wasp’ brand export quality rum. It follows the modern practices
of presentation of data in various board meetings. The data
___________________
collected by its Finance Director over a period of 3 years
___________________ pertaining to its operations is shown below.

E
___________________
1997- 1998- 1999-
Particulars
___________________ 1998 1999 2000

___________________ (a) Raw Material Cost per bottle of rum 9 15 21

UP
(b) Other costs 6 10 14
___________________
(c) Packing and Distribution 3 5 7
___________________
Sales proceeds per bottle (excluding excise) 20 30 40
___________________
Profit / (Loss) 2 0 (2)
___________________
Question

The Finance Director desires that the data should be presented

diagrammatically. Would you please help him in presenting the
data?
E-
Source: MBCQ-721D_Quantitative Techniques for Management Applications_CCE_UPES
CC
(c)
UNIT 21: Sampling

263

S
Notes

___________________

___________________

E
___________________

___________________

___________________

___________________

UP
___________________

___________________

___________________

___________________

BLOCK-V
E-
CC
(c)
Detailed Contents Quantitative Techniques for Management Applications

264

S
Notes
UNIT 21: SAMPLING
___________________ z Measuring Dispersion
z Introduction z Coefficient of Variance
___________________
z Sampling and Sampling Design
___________________ UNIT 24: FORECASTING

E
z Theory of Sampling
___________________ z Introduction
z Methods of Sampling
z The Conceptual Model
___________________
z Precautions in Using Sampling Methods
z The Mathematical Model
z
___________________
Sampling Reliability

UP
z Simple Forecasting System
z Sampling and Non-Sampling Errors
___________________
z Algorithms and Applications
___________________
UNIT 22: BASIC TOOLS OF DATA ANALYSIS–I z The Regression Algorithm
___________________
z Introduction z Simple Linear Regression Analysis
___________________
z Frequency Distributions z Correlation Analysis (The Parametric Case)
z Interpretation of Frequency Distributions z Correlation Analysis (The Non-parametric Case)
z Understanding and Using Averages z Time Series Analysis
z Kinds of Averages z Linear Analysis
E-
z Non-linear Analysis
UNIT 23: BASIC TOOLS OF DATA ANALYSIS–II
z Multiple Regression Analysis
z Introduction
z Commercial Averages z The Smoothing Algorithm

z Comparison of 3 M's of Statistics

UNIT 25: CASE STUDIES
CC
(c)
UNIT 21: Sampling

Unit 21
265

S
Activity
Notes
What is sampling means?

Sampling
___________________

___________________

E
___________________
Objectives
___________________
After completion of this unit, the students will be aware of the following
topics: ___________________

\ Sampling and Sampling Design ___________________

UP
\ Law of Statistical Regularity ___________________
\ Theory of Sampling
___________________
\ Sampling and Non-Sampling Errors
___________________

___________________
Introduction
If we want to make a study, which involves the total population,
there are two methods of doing it. One, we talk to each and every
member of the population and another which takes a
E-
representation of the whole population and does the study on it. It
is obvious that the second method is less costly, faster and easier
for us to use. The only problem that can crop up is that the
representation is not the true representation of the whole
population or is not a representation at all. For this there are
certain special statistical techniques used which help in checking
CC

that the representation used actually and truly represents the

population. These techniques are called sampling techniques and
the representation is called the sample.

Sampling and Sampling Design

Sampling can be defined as the selection of some part of an
aggregate or totality on the basis of which a judgment or inference
about the aggregate or totality is made. Thus, only after studying a
part of the whole population, inference is drawn on the whole
population. The whole population (or the desired group that we
want to study from which a sample is drawn) is called the universe.
(c)

Here onwards we will only use the term population.

Population could be finite or infinite depending upon the number of
elements in it. For example, the number of books that one
publisher sells is finite and can be known but the same thing
cannot be said about the number of people which have gone
 Quantitative Techniques for Management Applications

266 through these books so that becomes an infinite population.

S
Notes
Activity Population can also be divided into real and hypothetical. Real
What is sampling design?
___________________ population refers to hard facts which in the above case are the
number of books published. It can be hypothetical or imaginary, for
___________________
example, the number and types of emotions that you displayed in
___________________

E
last one hour! These can only be projected or imagined but you can
___________________ never be sure.
___________________ Characteristics and elementary units are the other two terms that
___________________ you need to know about here. Characteristics refer to the attributes

UP
(non-quantified qualities) which are the objects of the study.
___________________
Elementary units refer to those units which possesses the
___________________
characteristics of the population. The total of such elementary
___________________ units is called the population.
___________________ A sampling design is a definite plan for obtaining a sample from
the sampling frame. It refers to the technique or procedure of
selecting some sampling units from which inferences about the
population are drawn.
Sampling Errors: Since in sample survey, only a small part of
E-
the universe is studied, as such there is every possibility that its
result would differ to some extent from that of the universe. Even
if two or three samples of the same universe are taken, the result
would differ from each other. These differences constitute the
errors due to sampling and are known as sampling errors. Errors
due to calculations or improper convention of observation are
CC

called non-sampling errors.

Sampling Distribution: If we take certain number of samples
and for each sample and compute various statistical measures such
as mean, standard deviation, etc., then we can find that each
sample may give its own value for statistics under consideration.
All such values of a particular statistic, say mean, together with
their relative frequencies will constitute the sampling distribution.
The confidence level or reliability of the sample is found from the
sampling distribution. It is the expected percentage of times that
the actual value will fall within the stated precision limits. Thus, if
(c)

we take confidence level of 95%, we mean that there are 95

chances in 100 that sample results represent the true condition of
the universe. The significance level has the opposite interpretation
of that of confidence level and indicates the likelihood that the
answer will fall outside the range. You should always remember
that if the confidence level is 95% then the significance level will be
UNIT 21: Sampling

5%, which essentially means that there are five per cent chances 267

S
that the sample will not represent the true condition of the Notes
Activity
universe. The sample you have selected must follow the following Define law of statistical
___________________
law. regularity.
___________________

Law of Statistical Regularity

E
___________________

The law states that, if a moderately large number of items are ___________________
selected at random from a given universe, the characteristics of ___________________
those items will reflect, to a fairly accurate degree, the
___________________
characteristics of the entire universe. For example, if 500 leaves

UP
are picked from a tree at random and the average length is found ___________________

out, the result will be nearly the same if all the leaves of the tree ___________________
are picked up and measured.
___________________

The reliability in the Law of Statistical Regularity depends on two ___________________

factors:

(i) The larger the sample, the more reliable are its indications for
the population. The reliability of a sample is proportional to
E-
the square root of the number of items it contains and larger
the sample the more representative and stable it will be.

(ii) The sample must be chosen at random.

With the use of law we can say that a part of the population can
represent the population. When census is not possible due to
paucity of time, money and/or labour, then with the help of this
CC

law and random sampling, investigations can be made about the

properties of the population. This is possible because the selection
is made at random and by this law all types of units, whether good,
bad or average, have equal chance of being selected.

However, there are certain precautions which we should keep in

mind. The selection of units should be unbiased, i.e., random. The
two characteristics of randomization – One can not fit any
relationship between occurred values. Secondly, the probability of
occurrence of every item should be the same. The inferences drawn
from this are applicable on an all units of the population so the
(c)

sample should be identical to universe. By collecting information

from smaller units, we cannot apply the results drawn from it to
the whole universe.

The Law of Inertia of large numbers is a corollary of the Law of

Statistical Regularity. It lays down that in large masses of data,
 Quantitative Techniques for Management Applications

268 abnormalities will occur, but in all probability, exceptional items

S
Notes
Activity will offset each other, leaving the average unchanged subject to
What is theory of sampling?
___________________ where the elements of the time enters in the general trend of data.
The law of Inertia of large numbers asserts that large aggregates
___________________
are relatively more stable than small ones. The movements of an
___________________

E
aggregate are the result of the movements of its separate parts and
___________________ it is improbable that the later will be moving in the same direction
___________________ at the same time. Consequently, their movements will tend to
compensate one another and the larger the numbers involved, the
___________________
more complete will this compensation is. Thus, the law states that

UP
___________________ the larger the number of items we take out of a given universe, the
___________________ greater is the probability of accuracy.
___________________
Check Your Progress
___________________
Fill in the blanks:
1. The Law of Inertia of large numbers is a corollary of the
Law of ___________.
2. The reliability of a sample is proportional to the _______
E-
of the number of items it contains

Theory of Sampling
The theory of sampling is a study of relationships existing between
a universe and sample drawn from the universe. It is applicable
only to random sampling.
CC

The theory of sampling is concerned with estimating the properties

of universe from those of the sample and also with gauging the
precision of estimation. Sampling theory deals with the following
aspects:

(a) Statistical Estimation: Sampling theory helps in estimating

unknown population parameters from the knowledge of
statistical measures based on sample studies. The estimation
can be a point estimate or it may be an interval estimate.
Point estimate is a single estimate expressed in the form of a
single figure and interval estimate has two limits viz., the
(c)

upper limit and lower limit within which the parameter value
may lie. For example, we can say that the number of defective
parts in 100 pieces is 10 based on the sample of 10 in which 1
defective part was found or we can say that the defective parts
could be from 8-12 based on the many samples.
UNIT 21: Sampling

269
(b) Testing of Hypothesis: The second objective of the sampling

S
Notes
theory is to accept or reject the hypothesis. It helps in
determining whether observed differences are actually due to ___________________

chance or whether they are really significant. ___________________

E
(c) Statistical Inferences: Sampling theory helps in making ___________________

generalization about the universe from the studies based on ___________________

sample drawn from it. It also helps in determining, the
___________________
accuracy of such generalization.
___________________

UP
In the quantified research, the sampling technique is made
___________________
maximum use of and in no field of research can its importance and
___________________
value be belittled. In researches in the educational, economic,
commercial and scientific domains, the sampling technique is used ___________________
and considered most apt for research. Sampling technique also has ___________________
a very high value in day to day activities. In making our daily
purchases of foodstuff, vegetables, fruits, etc., it is not considered
necessary to examine each and every piece of the commodity. Only
a handful of goods are examined and the idea about the whole lot is
E-
formed and this usually proves justified. For example,
the physicians make inference about a patient’s blood through
examination of a single drop.

Sampling technique has the following features, which highlight its

importance:
CC

(a) Economy: The sampling technique is in expensive and less

time consuming than the census technique.

(b) Reliability: If the choice of sample units is made

with due care and the matter under survey is not
heterogeneous, the conclusion of the sample survey can have
almost the same reliability as those of census survey.

(c) Detailed study: Since the number of sample units is fairly

small, it can be studied intensively and elaborately. These can
be examined from multiple points of view.

(d) Scientific base: This is a scientific technique because the

(c)

conclusions are verifiable from other units. By taking random

samples we can determine the amount of deviation from the
norm.

(e) Greater suitability in most situations: Most of the surveys

are made by the technique of sample survey, because if matter
 Quantitative Techniques for Management Applications

270 is homogeneous, the examinations of few units suffice. This is

S
Notes the case in the majority of situations.
___________________
The question that arises is whether we can use sampling
___________________ techniques in any situation. No, sampling is useful only when:
___________________

E
(i) Data is vast: When the number of units is very large,
___________________ sampling technique must be used as it economizes money,
___________________ time and effort.

___________________ (ii) When utmost accuracy is not required: The sampling

UP
___________________ technique is very suitable in those situations where cent per
cent accuracy is not required otherwise census technique is
___________________
unavoidable, because 100% accuracy is achievable only by this
___________________ means.
___________________
(iii) Where census is not feasible: If we want to know the
mineral wealth in the country we cannot dig all the mines to
discover and count, we have to use the sampling technique.

(iv) Homogeneity: If all the units of a domain are alike, sampling

E-
technique is easier to use and is much more accurate.

The point to remember is that if due care is not taken in the

selection of samples or if they are arbitrarily selected, the
conclusions derived from them about the universe will be
misleading, if not totally wrong. For example, in assessing the
monthly expenditure of university students, if we select for our
CC

sample only the students who come in cars, our results will be
highly erroneous if extended to all students.

You should remember that the sampling technique can be

successful only if a competent and able investigator makes the
selection. If the sampling is done by an average investigator, the
selection may be prone to error.

Check Your Progress

Fill in the blanks:
1. The ___________ is in expensive and less time
(c)

consuming than the census technique.

2. Most of the surveys are made by the technique of
___________ .
UNIT 21: Sampling

Methods of Sampling 271

S
Notes
Activity
Sampling methods can be grouped under two broad categories: What is lottery method?
___________________
(i) Probability sampling methods or random sampling methods. ___________________

(ii) Non-probability sampling methods or non-random sampling

E
___________________
methods. ___________________

Probability Sampling Methods ___________________

Probability sampling methods are those in which every item in the ___________________

UP
universe has a known chance or probability of being included in ___________________
the sample. This implies that the selection of item for the sample is
___________________
independent of the person making the study and the items will be
___________________
chosen strictly at random.
___________________
Probability sampling can be divided into four types. We will take a
look at each one of them, one by one.

(i) Simple random sampling.

E-
(ii) Stratified sampling.

(iii) Systematic sampling.

(iv) Cluster sampling.

(i) Simple Random Sampling: A procedure of sampling will be

called simple random sampling where individual items (units)
CC

constituting the samples are selected at random. Random

sampling is the form applied when the method of selection
assures each individual element or unit in universe can have
an equal chance of being chosen. In other words, if in a sample
size of ‘n’ all the possible combinations of ‘n’ element items
have the same probability of being included; it is called simple
random sampling. It can be performed with replacement of
the taken out element or without replacement of the taken out
element.

Selecting a Random Sample

(c)

A random sample can generally be selected in following four

methods:

(a) Lottery method.

(b) Tippet’s numbers method.

 Quantitative Techniques for Management Applications

272
(c) Selection from sequential list.

S
Notes

___________________ (d) Grid system.

___________________ A brief description of the above methods is given below:
___________________

E
(a) Lottery method: In this method, a unit is drawn by writing
___________________ the numbers or the names of various units and putting
___________________ them in a container. They are thoroughly mixed and
certain numbers are picked up from the container, and
___________________
those picked up are taken up for sampling.

UP
___________________
(b) Tippet’s numbers method: It is called Tippet’s numbers
___________________
method because it was evolved by L.H.C. Tippet who
___________________
constructed a list of 10,4000 four digit numbers written at
___________________ random on every page. From those numbers it is not very
difficult to draw samples at random. For example, if 50
persons are to be selected for study out of the total number
of 500, then we can open any page of Tippet’s numbers and
select first 50 that are below 500 and take them up for
E-
study. On the basis of the experiments carried out through
this technique, it has been found that the results that are
drawn on the basis of this method of random sampling are
quite reliable.

(c) Selection from sequential list: In this method, the names

are arranged serially according to a particular order. The
CC

order may be alphabetical, geographical or only serial.

Then out of the list any number may be taken up.
Beginning of selection may be made from anywhere.

For example, if we want to select 10 persons, we can start

right from the 10th, and select 10, 20, 30, 40 and so on.

(d) Grid System: This method is generally used for selecting

the sample of an area and so in this method, a map of
entire area is drawn. After that a screen with the squares
is placed upon the map and some of the squares are
selected at random. Then screen is placed upon the map
(c)

and the areas falling within the selected squares are taken
as samples.

(ii) Stratified Sampling Method: In this method, the entire

population is divided into a number of groups called strata.
Then a number of items are taken from each group at random.
UNIT 21: Sampling

This means that a stratified sample is equivalent to a set of 273

S
random samples on a number of sub-populations. It can be Notes
performed with replacement of the taken out element or ___________________
without replacement of the taken out element.
___________________
As you would have guessed by now, in this method much

E
___________________
depends on the process of stratification. For taking the right
___________________
strata the following precautions need to be taken:
___________________
(a) Each stratum in the population should be large enough in
___________________
size so that selection of items may be done on random

UP
basis. ___________________

___________________
(b) There should be a perfect homogeneity in different units of
any one stratum. ___________________

___________________
(c) Stratification should be well defined and clear cut. By this
we mean that each unit or stratum should be free from
influence of the other.

Types of Stratified Sampling

E-
Stratified samplings are of three kinds:

(i) Proportionate stratified sampling: In this method the

number of units drawn from each stratum are in exact
proportion to proportion of strata to the population.

(ii) Disproportionate stratified sampling: In this type of

CC

stratified sampling an equal number of cases are taken

from each stratum without any consideration to the size of
strata in proportion to population. It is also called
“controlled sampling” or a quota is decided for each state.

(iii) Stratified weight sampling: In this method, an equal

number of units are selected from each stratum and
averages are drawn from each stratum, but in doing so
they are given weight in proportion to the size of stratum
in relation to the whole population.

(iii) Systematic Sampling: Systematic sampling is a variation of

(c)

simple random sampling. It requires that the universe or a

list of its units may be ordered in such a way that each
element of the universe can be uniquely identified by its order.
A voters list, a telephone directory, a card index system would
all generally satisfy this condition. Suppose, there are 5000
 Quantitative Techniques for Management Applications

274 cases (and hence 5000 units of the population) and we want a

S
Notes sample of 50. We can select a number of between (and
___________________ including) 1 and 10 at random, say 8. Then we can select the
units whose case is in the following position 18, 28, 38,.........,
___________________
1008,.........., 4998. This would be a systematic random sample
___________________

E
or commonly known as systematic sampling.
___________________
Systematic selection implies that the sample units are picked
___________________
out in a definite sequence, at equal intervals from one
___________________ another. Reduction or increase in the variability of estimates

UP
___________________ yielded by systematic sampling depends on the way
population is arranged. If the population is thoroughly mixed
___________________
with respect to the characteristics under study, the variability
___________________ of the estimates will be affected.
___________________
In practice, it is essential to use systematic sampling only
when we are sufficiently acquainted with the data to be able
to demonstrate that periodicities do not exist, or that the
interval between the elements of the sample is not multiple or
submultiple of the period.
E-
(iv) Cluster Sampling: Cluster sampling is also called
multistage sampling or sub-sampling as it uses various stages
to reach or make samples. This method is generally used in
selecting a sample from a very large population. The original
units into which the population is divided are known as
primary units. Each primary unit that falls into the sample is
CC

subdivided into secondary units in preparation for the second

stage of sampling. In three stage sampling; there will be
primary, secondary and tertiary units. Sometimes four stages
are also used.
Let us take an example to understand the procedure as it is
slightly tedious. Let us say that we want to take sample from the
universe of professors/lecturers associated with Delhi University.
The list consists of 100 pages with approximately 25 names per
page. These pages are numbered and constitute the sampling
units. All names are numbered and arranged alphabetically to
(c)

constitute the ultimate sampling units. Let us suppose we want a

sample of 100 professors/lecturers. The sample may be selected as
follows: we may decide to select 5 professors/lecturers each from 20
pages. Select a number from 1 to 5 at random say 3. Select pages 3,
8, 13, 28, ... and so on to 98. Then by the use of random numbers
UNIT 21: Sampling

select 5 names from each of 20 pages. This is a combination of 275

S
systematic and simple random sampling. Notes

The point to remember is that you should use a form of random ___________________
sampling in each of the sampling stages where there are two or ___________________
more than two stages.

E
___________________
The variability of estimates yielded by multistage sampling may be
___________________
greater than that of estimates yielded by simple random sampling
___________________
for equal size. The variability of estimates in multistage sampling
depends on the composition of primary units. There are three ___________________

UP
reasons for hesitating to recommend this method. ___________________

(i) The cost of the travel would be too high. ___________________

(ii) Control of the non-sampling errors would be difficult; and ___________________

(iii) A probability sample of small units drawn at one stage ___________________

requires a form which lists all the small units and such a plan
becomes costly by comparison.

Non-probability Sampling Methods

E-
Sampling methods which do not provide every element in the
universe a known chance of being included in the sample are
collectively known as non-probability sampling methods. Here the
selection process is partially subjective and does not use
randomization. In other words we use judgements based on
convenience and other considerations rather than probability
CC

considerations.

Non-probability sampling methods can be divided into basically

three groups:

(1) Judgement or purposive sampling,

(2) Convenience sampling, and

(3) Quota sampling.

Judgement or Purposive Sampling

In this technique the investigator has complete freedom in

(c)

choosing his sample according to his wishes and desires. Although

he will try his best to get the sample which is representative of the
population, his judgment plays a major part in determining which
the best sample is and no other considerations are used for the
same.
 Quantitative Techniques for Management Applications

276
When only a small number of sampling units are in the population,

S
Notes
simple random selection may miss the more important elements,
___________________
where judgement selection would certainly include them in a
___________________ sample. For example, when we want to know the effectiveness of
___________________ HR policies of the company and we randomly choose the sample of

E
___________________ 5 from a company of 50 people, it is possible that we only get

___________________
marketing people and not a representation of other functions in the
sample.
___________________

UP
___________________ This is bound to give us improper ideas about the effectiveness of
HR policies if the marketing people do not like the HR people in
___________________
the organization for whatever reasons. This personal selection can
___________________
become a disadvantage also when not used properly. Another
___________________ disadvantage that is associated with this method is that there is no
objective way of evaluating the reliability of sample results.

Still when we want to study some unknown traits of a population,

some of whose characteristics are known; we may then stratify the
E-
population according to these known properties and select
sampling units from each stratum on the basis of judgement. This
method will then result in a more representative sample.

Convenience Sampling

As the name suggests, in this method the sample of the population

CC

being investigated is selected neither by probability nor by

judgement but by convenience of reach. A sample obtained from
readily available lists such as automobile registrations, telephone
directories, etc., is a convenience sample and not a random sample
even if the sample is drawn at random from these lists. For
example, if you do a survey on the internet about the social issues,
your sample is a convenience sample and not representative of all
the people to whom these issues concern. Therefore, the results
obtained by convenience sampling methods are generally biased
and unsatisfactory.
(c)

So convenience sampling is normally suitable for doing pilot

studies and in cases where the population is not well defined or
sample units cannot be clearly defined or when the complete data
about the population is not available.
UNIT 21: Sampling

Quota Sampling 277

S
Notes
Quota sampling is a special form of stratified sampling. In this
___________________
method, first the classification of the population into various strata
is done in terms of properties known or assumed to be pertinent to ___________________
the characteristics being studied. Then proportion of the

E
___________________
population falling into each stratum on the basis of known or
___________________
estimated composition of the population is defined. After that the
quotas for each interviewer or investigator are determined so that ___________________

the total sample interviewed contains a proportion of each stratum ___________________

UP
so that all investigators study all the stratums thereby doing a
___________________
complete study of the population in a mini form.
___________________
The advantage of using this method is that items, which are close
___________________
to each other, are clubbed together, thereby saving costs and
___________________
introducing some stratification effect. The disadvantage is that the
bias of investigator is introduced in classification of subjects and in
random selection within various strata. Another disadvantage is
that since random sampling is not involved at any stage, the errors
of the method can not be estimated by statistical procedures.
E-
This method is mostly used in marketing surveys and election polls
and is pretty successful in that.

Multiphase Sampling
Sometimes it is economical and organizationally convenient also, to
CC

collect certain items of information from the all the units of a

sample and other items of information from some of these units
only, these latter units being selected so as to form a sub sample of
the original sample. This may be termed as two phase sampling or
double sampling. If necessary another phase may be added.

Multiphase or sequential sampling is of great use when the desired

accuracy of different items is widely different, either owing to the
fact that the variability of the associated variants is different or
because the desired accuracy is different.

The major advantage is that it is possible to select one sample from

(c)

a universe, analyze it and use the inferences in designing a second

sample from the same universe. But this method can only be used
where a small sample can represent the universe very well and
where the number of observations can be increased easily at any
stage of enquiry.
 Quantitative Techniques for Management Applications

278
Check Your Progress

S
Notes
Activity
Fill in the blanks:
What is multiphase sampling?
___________________

___________________
1. Quota sampling is a special form of _________ sampling.

___________________ 2. In ______________ method, the classification of the

E
population into various strata is done in terms of
___________________
properties known or assumed to be pertinent to the
___________________ characteristics being studied.
___________________
3. _________ or sequential sampling is of great use when

UP
___________________ the desired accuracy of different items is widely
___________________ different.

___________________
Precautions in Using Sampling Methods
___________________
In order to be useful, the study has to be representative in
character. If it does not possess all the characteristics of the
universe it shall not be representative enough and thus, it shall
not be able to fulfil the objective of the study. In order to enable the
E-
investigator to keep himself away from the biased samples, he has
to take the following precautions:

z If the universe is subject to change, an enquiry carried out on

a single occasion, howsoever accurate, cannot by itself give
any information on the nature of the rate of such change. In
such cases, provision must be made for studying the samples
CC

at successive intervals if up to date information is desired.

This successive study also gives some idea about the nature
and question of change.

z The size of the sample should not be too small as compared to

the universe. The size of the sample should be large enough so
that its representative character may not be lost and selection
on random basis may be possible. Apart from it, the sampling
should not be done purposely. In such an event, the sample
generally gets biased.

If the sampling is done through stratified method; it should

(c)

z
not be governed by the principle of perfect stratification.
Elements of unsuitability, overlapping or lack of proportion
have no place in sampling. When these elements are there,
the samples become biased.
UNIT 21: Sampling

279
z Lack of source list or incomplete source list makes the sample

S
Notes
biased.
___________________
z If the cases, which were originally selected for the study, are
___________________
lost or not available for enquiry, they are replaced by new

E
___________________
ones. In such a situation, there is a danger of the bias
influencing the selection of samples. ___________________

___________________
z When field workers are given the liberty to select samples
according to their wishes and no specific guidelines are given ___________________

UP
to them, they are likely to select samples according to their ___________________
convenience. In that event prejudices and bias are likely to ___________________
influence the sample.
___________________
z If the method of drawing samples is inadequate or not ___________________
suitable to the project, the samples drawn may be biased.
Sometimes the nature of the phenomenon makes the selection
of representative samples extremely difficult. It generally
happens in case of complex, heterogeneous and widespread
E-
cases.

z The investigator has to safeguard against the bias and try to

find out perfectly representative samples. In this, the
investigator and his skills play a vital role.

If the investigator is well equipped with the knowledge of the

CC

universe, know the importance and the nature of the study and
makes efforts to collect the representative sample, he would be
successful in selecting representative samples and take
precautions for removing the bias. Pre-testing is very helpful in
determining whether a particular sample is truly representative or
not.

Check Your Progress

Fill in the blanks:
1. ________ is very helpful in determining whether a
(c)

particular sample is truly representative or not.

2. Lack of source list or incomplete source list makes the
sample __________.
 Quantitative Techniques for Management Applications

280
Sampling Reliability

S
Notes

___________________
That the sample should be reliable and free from biases goes
without saying, but how that needs to be tested. The size of the
___________________
sample, its relevance and suitability to the problem, its
___________________ representation of the universe, etc., are some of the factors that

E
___________________ determine the reliability of the sample. Reliability may be tested
on the following parameters:
___________________

___________________ z Size of the sample: The size of the sample for study very

UP
much determines, not only its representative ness but also its
___________________
utility for study. The investigator must test that the size is
___________________ adequate for scientific and convenient study of the problem.
___________________
z By testing the representative ness of the sample: The
___________________ representativeness of the sample should also be tested. It
means that the sample selection should be representative and
possess the characteristics of other units.

z By drawing a parallel sample: It means that apart from

E-
the samples that have been drawn, another sample may be
drawn from the same universe for testing. On the basis of
these tests, the reliability of the sample, primarily selected,
may be tested. The comparison of two sample values gives you
a better understanding of sample.

z By testing the homogeneity of the samples: Samples

CC

should be homogeneous. They should possess all the

characteristics that are present in the population.

z Through comparison of the measurement of the sample

with those of the population: Sometimes, different
measurements about the universe are also known. In order to
test the reliability of the samples, the investigator may apply
his knowledge and thereby test the reliability of the sample.

z Unbiased selection: The selection of sample should be done

through a method that is free from bias and prejudices.

By taking a sub-sample from a main sample: This is a

(c)

„
method of sampling within sampling. Out of the universe, we
draw a sample, but in order to test the reliability of the
sample, we draw a sub-sample, from main sample and study it
intensively and compare the findings of the study of the sub-
sample with the findings of the study of the main sample. This
UNIT 21: Sampling

helps the investigator to detect any error that might have 281

S
crept in. Notes
Activity
What are sampling errors?
___________________
Check Your Progress
___________________
Fill in the blanks:

E
___________________
1. The size of the sample for study very much determines
___________________
not only its representativeness but also its _________ for
study. ___________________

2. The _______________must test that the size is adequate ___________________

UP
for scientific and convenient study of the problem. ___________________

___________________
Sampling and Non-Sampling Errors ___________________
Errors in any statistical investigation, i.e., in collection, processing ___________________
and analysis of the data, may be broadly classified as: (i) Sampling
errors and (ii) Non-sampling errors.

Sampling Errors
E-
In a sample survey, only a small part of the universe is studied. As
such there is every possibility that its results would differ to some
extent from that of the universe. Even if two or more samples of
the same universe are taken the results will differ to some extent
from that of the universe, as well as from each other. The
difference would be always present even if the sample is drawn at
CC

random. These differences constitute the errors due to sampling

and are known as sampling errors. This is the error which is the
result of sample or sampling procedural and it always exists in
same quantity.

Sampling errors creep in because of the following reasons:

(i) Faulty selection of the sample: Purposive selection of

sample would result in biases, as in this case, the investigator
deliberately selects the representative sample. If, however,
the selection of sample is haphazard, the chances of bias
errors are great.
(c)

(ii) Incomplete investigation (or non-response): If all the

items to be included in the sample are not covered there will
be bias. This occurs frequently in case of sample collected
through questionnaire method. All the questions in the
questionnaires are not responded properly. Again if the
 Quantitative Techniques for Management Applications

282 selected person is not interviewed at any time to collect

S
Notes information, bias may arise.
___________________
(iii) Faulty collection of data: During the process of collection of
___________________
data certain errors and mistakes may creep in due to the
___________________ following reasons:

E
___________________
(a) Negligence or prejudice of enumerator in putting questions
___________________ or recording answers.
___________________
(b) Lack of knowledge on the part of person furnishing

UP
___________________ information.
___________________
(c) Poorly designed questionnaire.
___________________
(d) Unorganized method of collection of data.
___________________

(iv) Substitution: Due to non-availability of the selected person

(or item) the investigator may interview another person from
the same sample. The second person may not have the same
characteristics as the original one. This will introduce the
E-
substitution bias in the sample and as such deviate the result.

(v) Faulty Analysis: Faulty method of analysis of data may also

introduce the sampling error.

Enough has been written about the biases of the investigator or

the respondents. There could also be unbiased errors that creep in
CC

due to accident or by natural course of events without any bias by

enumerator or informant. They occur due to chance factor that of
the member of universe being excluded or included in the sample
selection. Further, this type of error occurs when only a partial
observation of universe is made and is equal to the difference
between sample statistic and parameter of universe.

If the overestimated and underestimated values of observations

are nearly equal, then errors in one direction will compensate the
errors of other direction. Therefore, the unbiased errors are usually
known as compensatory errors as they tend to offset each other
(c)

and leave little effect on the general results.

Non-sampling Errors
The above discussion about the sampling errors seems to imply
that studies of the entire population are free from any errors.
UNIT 21: Sampling

Nothing could be farther from the truth. Errors may occur at any 283

S
stage of enquiry, i.e., planning, collection, processing and analysis Notes
of data. Apart from sampling errors, errors arise due to following ___________________
reasons:
___________________
z Faulty planning resulting in improper definition of the

E
___________________
problem statement.
___________________
z Vague or incomplete definition of universe.
___________________
z Imperfect questionnaire which might result in incomplete or
___________________
wrong information.

UP
___________________
z Defective data collection.
___________________
z Acceptance of exaggerated or irrelevant or wrong answers to
___________________
the questions that satisfy the pride or self-interests of the
respondents. ___________________

z Personal bias of the user of the report.

z Improper understanding and definition of the variables.

E-
z Improper use of averages to replace the actual figures.

z Application of wrong methods.

z Defect in measuring instruments.

And this list is by no means exhaustive!

The magnitude of the errors that can creep in indicate
CC

that sampling is a technique which must be used selectively and

objectively. Reducing sampling and non-sampling
errors can be achieved by keeping in mind the above mentioned
points and not loosing the sight of the objectives of the study at any
stage. Still some errors would invariably creep in.

Check Your Progress

Fill in the blanks:
1. There could also be ___________ errors that creep in due
to accident or by natural course of events without any
bias by enumerator or informant.
(c)

2. Selection of the ___________ problem, as has already

been stated, should be in line with the researcher's
interest, chain of thinking and existing research in the
same area and should have some direct utility.
Contd…
 Quantitative Techniques for Management Applications

284
3. What is most important in selecting a research problem

S
Notes
is that the research topic should be within ___________ .
___________________
4. Grid system is applied for selection of sample from a
___________________
___________ .
___________________

E
5. Purposive selection of samples, as the name goes,
___________________ depends more on the researcher's ___________ .
___________________
6. Stratified sampling combines the characteristics of
___________________ random sampling and ___________ .

UP
___________________

___________________ Summary
___________________ Sampling can be defined as the selection of some part of an
___________________
aggregate or totality on the basis of which a judgment or inference
about the aggregate or totality is made. Thus, only after studying a
part of the whole population, inference is drawn on the whole
population. Population could be finite or infinite depending upon
the number of elements in it.
E-
A sampling design is a definite plan for obtaining a sample from
the sampling frame. It refers to the technique or procedure of
selecting some sampling units from which inferences about the
population are drawn. The law states that, if a moderately large
number of items are selected at random from a given universe, the
characteristics of those items will reflect, to a fairly accurate
CC

degree, the characteristics of the entire universe. The theory of

sampling is concerned with estimating the properties of universe
from those of the sample and also with gauging the precision of
estimation.

Sampling methods can be grouped under two broad categories: (i)

Probability sampling methods or random sampling methods and
(ii) Non-probability sampling methods or non-random sampling
methods.

Errors in any statistical investigation, i.e., in collection, processing

and analysis of the data, may be broadly classified as: (i) Sampling
(c)

errors and (ii) Non-sampling errors.

Lesson End Activity

Write a note on non-probability sampling methods.
UNIT 21: Sampling

Keywords 285

S
Notes
Sampling: Sampling can be defined as the selection of some part
___________________
of an aggregate or totality on the basis of which a judgment or
inference about the aggregate or totality is made. ___________________

Sampling Design: A sampling design is a definite plan for

E
___________________
obtaining a sample from the sampling frame. ___________________

Simple Random Sampling: A procedure of sampling will be ___________________

called simple random sampling where individual items (units)
___________________

UP
constituting the samples are selected at random.
___________________
Non-sampling Errors: Errors due to calculations or improper
___________________
convention of observation are called non-sampling errors.
___________________

Questions for Discussion ___________________

1. Define sampling.
2. What is the law of statistical regulatory?
3. Write a note on sampling theory.
E-
4. What are the key features of sampling techniques?
5. What are the key methods of sampling?
6. Make a difference between simple random sampling and
stratified sampling.
7. Differentiate between the sampling and non-sampling errors.
CC

Further Readings

Books
R S Bhardwaj, Mathematics for Economics and Business, Excel
Books, New Delhi, 2005

D C Sanchethi and V K Kapoor, Business Mathematics

Sivayya and Sathya Rao, An Introduction to Business Mathematics

Web Readings
(c)

http://www.math.ucla.edu

www.textbooksonline.tn.nic.in

www.mathbusiness.com
 Quantitative Techniques for Management Applications

286

S
Notes

___________________

___________________

___________________

E
___________________

___________________

___________________

UP
___________________

___________________

___________________

___________________
E-
CC
(c)
UNIT 22: Basic Tools of Data Analysis–I

Unit 22
287

S
Activity
Notes
What is the difference

Basic Tools of Data Analysis–I

___________________
between collecting and
collating data?
___________________

E
___________________
Objectives
___________________
After completion of this unit, the students will be aware of the following
topics: ___________________

\ Frequency Distributions ___________________

UP
\ Understanding and Using Averages ___________________
\ Kinds of Averages
___________________

___________________
Introduction
___________________
Collecting and collating data is one serious matter as we saw in the
last unit, getting a meaningful analysis out of it is another. In the
next few units we will focus on understanding the tools that we
require to analyze this data.
E-
The data collected could be in terms of qualitative variables or in
terms of quantitative variables. Examples of qualitative variables
include items termed as defective or effective, persons classified as
rich or poor, etc. Care must be taken before we quantify these
qualitative variables for this is one of the major sources of errors.
Quantitative variables may be discrete, continuous or a
CC

combination of the two. Discrete variables take on only whole

number values, for example, number of defective parts in a sample,
number of married people in a city, etc. Continuous variables can
be measured to any arbitrary degree of accuracy, for example, the
weight of a person can be measured to the nearest kilograms,
grams, milligrams, etc. The result may or may not be a whole
number. The accuracy desired should be such that the relevancy of
the data is not lost and it is not too difficult to get the desired data.

Frequency Distributions
(c)

The data that you have collected till now, either through sampling
or otherwise, are called raw data. Now this data can be arranged in
an array. For example, if you collected data on electricity
consumption for one day of 1000 households, you would get an
array with 1000 rows and two columns, 100 rows for the houses
and the two columns for house numbers and electricity
 Quantitative Techniques for Management Applications

288
consumption respectively. As it is very difficult to draw inference

S
Notes from this raw data, we can process this data so as to show the
___________________ number of houses, which are using electricity within a particular
range, together. This table of electricity consumption ranges and
___________________
number of houses is shown below:
___________________

E
Table 22.1 Frequency Distribution of Electricity Consumption
___________________
Electricity Consumption (kilowatts) Number of houses
___________________ 0-9 1
___________________ 10-19 3

UP
20-29 5
___________________
30-39 10
___________________
40-49 20
___________________ 50-59 35

___________________ 60-69 50
70-79 70
80-89 100
90-99 130
100-109 130
E-
110-119 100
120-129 70
130-139 50
140-149 35
150-159 20
160-169 10
170-179 5
CC

180-189 3
190-199 1

Note that, we are not using the house numbers as the information
is irrelevant for the purpose of determining how many houses use
how much electricity per day. Also note that, this is only one day’s
utilization and is not indicative of the average utilization of
electricity by that household. If it is Sunday, the overall average
would be lower that what you can draw an inference from this
data. This kind of over-generalization is very common and is a
frequent source of errors in real life situations. It is important that
(c)

a frequency distribution should have a suitable number of class

intervals. Class intervals mean the ranges for which we classified
the number of items. In the above case 10 is the class interval used
for electricity consumption. If too few classes are used, the original
data would be so compressed that little information will be
available. If too many classes are used, there will be too few items
UNIT 22: Basic Tools of Data Analysis–I

in the classes, and the frequency polygon would be irregular in 289

S
appearance. Notes
Activity
There are basically three precautions that must be kept in mind Define the term average
___________________
when determining the class intervals. First we must select the ___________________
class interval so that the mid-values of the classes will coincide, as

E
___________________
far as possible, with the concentration of items that may be
present. Second, we should avoid open-ended classes. Third, the ___________________

class intervals should usually be uniform. ___________________

___________________
Check Your Progress

UP
___________________
Fill in the blanks:
___________________
1. The data that you have collected till now, either through
sampling or otherwise, is called ________ data ___________________

___________________
2. There are basically three precautions that must be kept
in mind when determining the class intervals. First
___________ Second, ______________ Third, ___________.
E-
Interpretation of Frequency Distributions
Frequency distributions may differ in average value, dispersion,
shape or any combination of the three. Mathematicians have
named almost all kinds of shapes which combine the properties of
these three variations. They do not add too much value to your
learning and therefore, are not mentioned here.
CC

We have discussed averages, dispersion and skewness graphically

and you must be thinking that there must be a way to measure
these quantities mathematically. There is and that is what we are
going to discuss next before going on to probability distributions.

Understanding and Using Averages

In a series of statistical data that parameter which reflects a
central value of the series is called the central tendency. Central
tendency refers to the middle point of a statistical distribution and
is also known as an average.
(c)

Average as a Measure of Central Tendency

An average can be defined as a central value around which other
values of series tend to cluster. An average is computed to give a
concise picture of a large group: By the use of average complex
 Quantitative Techniques for Management Applications

290 groups, large numbers are presented in a few significant words or

S
Notes figures. Averages help in obtaining a picture of the universe with
___________________ the help of sample. Although sample and the universe differ in
size, still their average may be very much identical.
___________________

___________________ Averages give a mathematical concept to the relationship between

E
different groups, for example, the trees in one forest are taller than
___________________
in another forest but in order to find any definite ratio of heights it
___________________
is essential to resort to averages.
___________________
But is an average a representative? Yes, essentially because of

UP
___________________ three reasons:
___________________
(i) Ordinarily most of the values of a series cluster in the middle,
___________________
(ii) At the extreme ends the number of items is usually very little,
___________________
and

(iii) Ordinarily items with values less than the average cancel out
the items whose values are greater than the average – The
average of 4, 5, 6 is 5. The average 5 is less in value and is
E-
more in value by one towards both the extremes. Thus, the
two deviations -1 and +1 cancel each other.

An average should be affected as little as possible by sampling

fluctuations, i.e., for different sample of same population the
variation in the average is very little. An average should be
capable of algebraic treatment so that it can be used for further
CC

mathematical manipulation.

Check Your Progress

Fill in the blanks:
1. An average should be affected as little as possible by
sampling ____________.
2. For different sample of same population the variation in
the average is ________________

Kinds of Averages
(c)

Averages may be classified into three broad types:

1. Mathematical Averages:

(a) Arithmetic mean

UNIT 22: Basic Tools of Data Analysis–I

291
(b) Geometric mean

S
Notes
Activity
(c) Harmonic average Establish the relation between
___________________
mean, median and mode.
2. Positional Averages: ___________________

(a) Mode

E
___________________

(b) Median ___________________

___________________
3. Commercial Averages:
___________________
(a) Moving average

UP
___________________
(b) Progressive average
___________________
(c) Quadratic average
___________________
Mathematical averages are those which utilize mathematical
___________________
formula for the calculation of their values. Positional averages do
not use mathematical calculations but give you an indication about
the positional characteristics of certain items. Commercial
averages are the applications of averages in commercial situations.
E-
If so many varieties of averages are there, the question that arises
is which one to use. As we go ahead we would see that each type
has a specific application and should be used only in that case.

Mathematical Averages

Arithmetic Mean
CC

Most of the time when we refer to the average we are talking about
arithmetic mean. This is true in cases like average winter
temperature in Delhi; average life of a flash light battery, average
working hours of an executive, etc. The arithmetic mean (or simply
mean) is the quantity obtained by dividing the sum of the values of
items (∑X) in a variable by their number (n), i.e., number of items.

ΣX
X=
n

The mean of 3, 4, 5, 6, 7 is.

(c)

ΣX 3 + 4 + 5 + 6 + 7
X= = =5
n 5
Looking at the formula above we can say that the algebraic sum of
the deviations of the individual items from the arithmetic mean is
zero. If the sum of the deviations of individual items from the
 Quantitative Techniques for Management Applications

292 mean is zero, then the sum of squares of the deviations is

S
Notes minimum when taken from the arithmetic mean than taken from
___________________ any other item.
___________________ This means that if any one or more items in the group are replaced
___________________ by new items, the new arithmetic mean would be changed by the

E
net change divided by number of items. For example, if the values
___________________
3 and 4 in the above example changes to 8 and 9 (total change of
___________________ 10) then the mean can be calculated in either of the two ways
___________________ mentioned below:

UP
___________________ ΣX 3 + 4 + 5 + 6 + 7
X= = =7
___________________
n 5

___________________ OR

Change in Value 10
___________________ New X = Old X + =5+ =7
n 5
Although in this example it would have been faster to do it the
original way, the alternate method assumes more and more
significance as the number of items go up.
E-
When a frequency distribution is given, as in the table 22.1, the
mean is calculated using a variation on the above formula.

Σfx
X=
n

Here f stands for frequency of the class, x stands for mid-value of

CC

the class and n stands for total of all frequencies in all classes.

Revised Table 22.1 is reproduced below as Table 22.2.

Table 22.2: Frequency distribution of
Electricity Consumption

Electricity Consumption Number of

Mid-value (x) fx
(kilowatts) houses (f)
0-9 5 1 5
10-19 15 3 45
20-29 25 5 125
30-39 35 10 350
40-49 45 20 900
(c)

50-59 55 35 1925
60-69 65 50 3250
70-79 75 70 5250
80-89 85 100 8500
Contd…
UNIT 22: Basic Tools of Data Analysis–I

90-99 95 130 12350 293

S
100-109 105 130 13650 Notes
110-119 115 100 11500
___________________
120-129 125 70 8750
130-139 135 50 6750 ___________________
140-149 145 35 5075

E
___________________
150-159 155 20 3100
___________________
160-169 165 10 1650
170-179 175 5 875 ___________________
180-189 185 3 555 ___________________

UP
190-199 195 1 195
___________________
Total Value 848 84800
___________________
Applying this formula to the table we get:
___________________

Σfx 84800 ___________________

X= = = 100
n 848

In the case where cumulative percentage distribution is given,

grouped frequency distribution is derived from the cumulative
percentage distribution and then the usual procedure is applied for
E-
computing the mean.
If two or more groups contain respectively N1, N2, … observations
with means X 1, X 2, … respectively, then the combined mean (X) of
the composite group is given by the relation:

NX1 + NX 2 + NX3 + ...

CC

XN =
N1 + N 2 + N 3 + ...

Here N stands for the sum of the denominator (N1+N2+N3+…)

Table22.3: Critical Evaluation of Arithmetic Mean

Merits Demerits
z It is rigidly defined and is z When distribution is highly skewed on
definite either side, arithmetic mean looses its
representativeness
z Its calculation is easy and z Its calculation requires information
generally understood about all units, either individually or
collectively. Therefore, it can not be
safely used in open end tables
z The data needs very little z It is not suitable for non-homogeneous
(c)

preparation, e.g., it need not series

be arrayed
z It utilises all the data in the z Can't be applied for extremely large
groups values on either side
z It is suitable for arithmetic
and algebraic manipulation
 Quantitative Techniques for Management Applications

294 Weighted Average

S
Notes
In calculating simple arithmetic mean it is assumed that all items
___________________
were equal in importance. It may not be the case always. When
___________________ items vary in importance they should be assigned weights in order
___________________ of their relative importance. For calculating the weighted

E
arithmetic mean the value of each item is multiplied by its weight,
___________________
product summated and divided by the total of weights and not by
___________________ the number of items. The result is the weighted arithmetic
___________________ average. Symbolically:

UP
x1 w1 + x 2 w2 + x 3 w3 + ...
___________________ Xw =
w1 + w2 + w3 + ...
___________________

___________________ Here w1, w2, w3, … stands for the respective weights of each of the

___________________ items.

Weighted averages have important applications in trend analysis

and forecasting. But it should be used when any of the following
conditions holds true:
E-
1. When the importance of all the items in a series is not equal.

2. When the items falling into different grades of the classes of

the same group show considerable variation and it is desired
to obtain an average which would be representative of the
whole group, weighted average is the only proper average to
be used. In other words, when the classes of the same group
CC

contain widely varying frequencies.

3. When the percentages, rates or ratios are being averaged.

4. When there is a change either in the proportion of frequencies

of items or in the proportion of their values.

It would not be improper to remind you that simple mean and

weighted mean are two of the mostly used means.

Geometric Mean
The most important application of geometric mean is in the
(c)

construction of index numbers, i.e., averaging rates of change. For

example, if you are investing in the stock markets, and your money
grows from ` 1,00,000 to ` 2,50,000 in three years and you want to
know what is the average percentage gain you are making over the
three years, you can use this mean.
UNIT 22: Basic Tools of Data Analysis–I

295
This could simply be written as:

S
Notes
X3
G= 3 ___________________
X0
___________________
This leads us to the general formula for geometric mean

E
___________________
Xn
G= n ___________________
X0
___________________
Mathematically speaking, the geometric mean is the nth root of the ___________________

UP
product of n items of a series.
___________________
The only problem with using this formula is that, you cannot do it
___________________
on a simple calculator and this is the biggest drawback of it.
___________________
Geometric mean is also useful in skewed distributions and
averaging ratios. ___________________

Table 22.4: Critical Evaluation of G.M.

Merits Demerits
z It gives comparatively little weight z It cannot be used when any of the
E-
to extreme values. quantities is zero or negative

z It is suitable for arithmetic and z It is difficult to compute and

algebraic manipulation requires more time in
computation
z It is reversible both ways and z It is difficult to understand
therefore, suitable for ratios and
percentages
CC

z It gives less weight to large items

which sometimes may be a
limitation, viz., computing
average cost per unit.

Harmonic Mean

It is defined as the reciprocal of the arithmetic mean of the

reciprocals. Thus, for a simple harmonic mean:
n
H=
1 1 1 1 1
+ + + + ... +
x1 x 2 x 3 x 4 xn
(c)

For a weighted harmonic mean, the above equation is rewritten as:

Σw
Hw =
⎛1⎞ ⎛1 ⎞ ⎛1 ⎞ ⎛1 ⎞ ⎛1 ⎞
w1 ⎜ ⎟ + w2 ⎜ ⎟ + w3 ⎜ ⎟ + w4 ⎜ ⎟ + ....wn ⎜ ⎟
⎝ x1 ⎠ ⎝ x2 ⎠ ⎝ x3 ⎠ ⎝ x4 ⎠ ⎝ xn ⎠
 Quantitative Techniques for Management Applications

296
Although harmonic mean is of limited use, it is less affected by

S
Notes
extremely large observations than any other average. It is properly
___________________
used to average rates where the weights are the numerators of the
___________________ fractions used to compute the rates.
___________________
Positional Averages

E
___________________
Mode
___________________

___________________
Mode is that value which has the maximum frequency (i.e. occurs
most often) in a given set of values. Thus the mode of a set of data

UP
___________________ is simply the value that is repeated most often. It is the most
___________________ typical value and, therefore, the clearest example of a measure of
central tendency.
___________________
For example, if you leave for your office everyday in the morning
___________________
and you recorded the following times for two weeks:
8.30 8.25 8.35 8.29 8.31 8.30 8.32 8.31 8.31 8.31
One thing is obvious, you are quite punctual! Anyway if you
arrange the data in increasing order
E-
8.25 8.29 8.30 8.30 8.31 8.31 8.31 8.31 8.32 8.35
Here 8.31 occurs most frequently and is therefore the mode of the
given range.
You must be thinking that there usually be two items of exactly
the same size for a continuous variable, (if measurements are
made with sufficient precision), it is apparent that our definition of
CC

the mode is somewhat vague. For this we group the data and then
use this simple equation:
d1
Mode = l1 + × C, where
d1 + d 2

l1 = lower boundary of the class containing the largest frequency

d1 = difference of the largest frequency and the frequency of the
last class
d2 = difference of the largest frequency and the frequency of the
next class
C = class interval
(c)

The main advantage of mode is that the value of mode is not

affected by the extreme values of the series. Plotting also is not
difficult, if there are more than one mode in the series, then it is
not difficult to determine it and it can be located graphically also,
very easily.
UNIT 22: Basic Tools of Data Analysis–I

297
Table 22.5: Critical Evaluation of Mode

S
Notes
Activity
Merits Demerits
What are grouped and
___________________
z It is very easy to locate. In many cases z It is frequently ill defined
ungrouped distribution?
it can be obtained by inspection. ___________________
z It is not influenced by the presence of a z It sometimes is

E
small number of extreme items. indeterminable without ___________________
modifying the data
___________________
z It may be ascertained even when the z It cannot be calculated by
details of extreme items are not simple arithmetic process ___________________
available.
z It is easily understandable. z It is unsuitable for arithmetic ___________________

UP
and algebraic manipulation
___________________
z It may be determined with
considerable accuracy from a well ___________________
selected sample data.
___________________
Median ___________________

Median is the value of that item in a series which divides the

series into two equal parts, one part consisting of all values less
and the other all value greater than it. Defined in another way
median is that value of the central tendency, which divides the
E-
total frequency into two halves.
Table 22.6: Critical Evaluation of Median

Merits Demerits
It is easy to understand It cannot be calculated by mathematical
and calculate methods and therefore is not suitable for
algebraic treatment
CC

It eliminates the effect of Median is usually affected by fluctuations of

extreme items sampling. Data must be arranged before
calculation of median
In many cases it may be It is unsuitable when greater importance to
obtained by inspection large or small values is to be given
It is easy to locate, subject It may be difficult to locate and when located it
to the actual number of may not be representative in case the items in a
items being known series are not suited closely together
Its position is more definite A correct total cannot be obtained by
than that of mode multiplying the median by the number of items.
It is clearly and rigidly
defined.

Calculation of median from simple series is very simple. If the data

(c)

set contains an odd number of items, the middle item of the array
after arrangement is the median. If there is an even number of
items the median is the average of the two middle items after
arrangement.
 Quantitative Techniques for Management Applications

298
Calculation of median from simple frequency distribution

S
Notes
(Ungrouped) is also easy. The cumulative frequency (less than
___________________
type) corresponding to each distinct value of the variable is
___________________ calculated. If the total frequency is N, the value of the variable
___________________ corresponding to cumulative frequency gives the median.

E
___________________ Calculation of median from simple frequency distribution
(Grouped) is slightly more complex. The following formula is used:
___________________

___________________ N
−F

UP
___________________ Median = L+ 2 ×C
Fm
___________________
Where,
___________________
L = Lower boundary of the median class
___________________
N = Total frequency

F = Cumulative frequency below the class immediately preceding

the median class
E-
Fm = Frequency of the median class

C = Class interval or width of the median class.

Check Your Progress

Fill in the blanks:
1. Median is the value of that item in a series which
CC

divides the series into two ________ parts.

2. Mode is that value which has the __________ frequency
(i.e. occurs most often) in a given set of values

Summary
The data collected could be in terms of qualitative variables or in
terms of quantitative variables. The data that you have collected
till now, either through sampling or otherwise, are called raw data.
As it is very difficult to draw inference from this raw data, we can
process this data so as to show the number of houses, which are
(c)

using electricity within a particular range, together.

In a series of statistical data that parameter which reflects a
central value of the series is called the central tendency. Central
tendency refers to the middle point of a statistical distribution and
is also known as an average.
UNIT 22: Basic Tools of Data Analysis–I

299
An average can be defined as a central value around which other

S
Notes
values of series tend to cluster. Averages may be classified into
three broad types: Mathematical Average, Positional Averages and ___________________
Commercial Average. ___________________

E
___________________
Lesson End Activity
___________________
What are the key situations in which weighted average is more ___________________
suitable than other methods of computing averages?
___________________

UP
___________________
Keywords
___________________
Average: An average can be defined as a central value around
___________________
which other values of series tend to cluster.
___________________
Mode: Mode is that value which has the maximum frequency (i.e.
occurs most often) in a given set of values.

Median: Median is the value of that item in a series which divides

the series into two equal parts, one part consisting of all values
E-
less and the other all value greater than it.

Mathematical Averages: Mathematical averages are those which

utilize mathematical formula for the calculation of their values.

Positional Averages: Positional averages do not use

mathematical calculations but give you an indication about the
CC

positional characteristics of certain items.

Moving Average: The moving average is an arithmetic average of

data over a period and is updated regularly by replacing the first
item in the average by the new item as it comes in.

Questions for Discussion

1. Illustrate the preparation of frequency distribution with a
suitable example.

2. Define average. What are the different types of averages and

(c)

their implication?

3. What are the different types of positional average?

 Quantitative Techniques for Management Applications

300
Further Readings

S
Notes

___________________ Books
___________________ R S Bhardwaj, Mathematics for Economics and Business, Excel
___________________ Books, New Delhi, 2005

E
___________________ D C Sanchethi and V K Kapoor, Business Mathematics
___________________ Sivayya and Sathya Rao, An Introduction to Business Mathematics
___________________
Web Readings

UP
___________________
http://www.math.ucla.edu
___________________

___________________ www.textbooksonline.tn.nic.in

___________________ www.mathbusiness.com
E-
CC
(c)
UNIT 23: Basic Tools of Data Analysis–II

Unit 23
301

S
Activity
Notes
What is meant by commercial

Basic Tools of Data Analysis–II

___________________
average?
___________________

E
___________________
Objectives
___________________
After completion of this unit, the students will be aware of the following
topics: ___________________

\ Commercial Averages ___________________

UP
\ Comparison of 3 M's of Statistics ___________________
\ Measuring Dispersion
___________________

___________________
Introduction
___________________
In this unit, we will discuss about commercial averages which will
include moving average, progressive average and quadratic
average. We will further study comparison of the mean, median
and mode. We will also focus on measuring dispersion.
E-
Commercial Averages
These are classified as follows:

Moving Average
CC

The moving average is an arithmetic average of data over a period

and is updated regularly by replacing the first item in the average
by the new item as it comes in. It is useful in eliminating the
irregularity of time series and is generally computed to study the
trend.
Suppose the prices for 12 months are given and a three monthly
average is to be computed. Then the first item in the 3-month
moving average would be the average [(a1+a2+a3)/3], the second
item would be the average of the next three months [(a2+a3+a4)/3]
and so on. The last item would be the average [(a10+a11+a12)/3]. As
the next month would come in a10 would be dropped and a13 would
(c)

be added in [(a11+a12+a13)/3] and so on.

Progressive Average
Progressive average is also calculated with the help of simple
arithmetic mean. It is a cumulative average. In computation of
 Quantitative Techniques for Management Applications

302
progressive average, figures of all previous years are added and

S
Notes
Activity divided by the number of items. As the number of items go up and
Give___________________
an example of reach a desired number, we switch to moving average.
progressive average.
___________________ Quadratic Average
___________________ The quadratic mean or average is estimated by taking the square

E
___________________ root of the average squares of the items of a series.
___________________ Symbolically,
___________________
a2 + b2 + c2 + ... + n 2

UP
Qm =
___________________ N
___________________ Where Qm = Quadratic Mean
___________________
a2, b2, c2 .....n2 = squares of the different values
___________________
Quadratic average is useful when some items have negative values
and other positive values because in such cases the mean is not
very representative. It is also used in averaging deviations, rather
than original values, when the standard deviation is computed.
E-
Check Your Progress
Fill in the blanks:
1. The moving average is an _________ average of data
over a period.
2. Quadratic average is useful when some items have
CC

______________ values and other positive values because

in such cases the mean is not very representative.

Comparison of 3 M's of Statistics

A comparison of the mean, median and mode is necessary for you
to understand the positive and negative characteristics of them.
Table 23.1: Comparison of 3 M's of Statistics

Parameter of Comparison
Average Mean is a calculated average. Median and Mode are
averages of positions. If all the items in a variable are the
(c)

same, then only AM=GM=HM otherwise AM > GM > HM

Calculation Mean is the sum of the values of the items divided by the
number of items in the series. Median is the middle value
which divides a series into two equal parts. Mode is the
value around which the items of a series tend to
concentrate in density, i.e., most occurred frequency.
Contd….
UNIT 23: Basic Tools of Data Analysis–II

Treatment Mean is capable of mathematical treatment. Median and 303

S
Mode are not capable of such treatment.
Notes
Activity
Items All the items in a series are taken into account in the
What are 3M’s of statistics?
calculation of Mean. Median and Mode calculations do ___________________
not consider all the items in a series.
___________________
Extreme Values Mean is affected by extreme values of the items in a
series but it is not so in case of Median and Mode.

E
___________________
Calculation in Mean calculations of frequency distribution with open-
ended ___________________

Open-ended class intervals at both ends is not possible. Median and ___________________
distributions mode of such distribution can he easily calculated.
Reliability Mean is considered to be more reliable measures of ___________________

UP
central tendency than Median and Mode.
___________________
Result In a series of distribution of data, there is only one value ___________________
of mean or median. But there could be more than one
mode or no mode at all. Mean is simple to understand ___________________
and to calculate.
Use Mean is widely used. Median and Mode have limited use. ___________________

Check Your Progress

Fill in the blanks:
E-
1. _____ is a calculated average.
2. Median and Mode are averages of __________

Measuring Dispersion
It is only because of variability that we compute averages. But if
CC

there is too much variability among the data, an average is so

unreliable that it is almost useless. Usually, a high degree of
uniformity (i.e. a small amount of dispersion) is a desirable quality.
Mass production would usually be uneconomical if there was a
large amount of variability in materials or manufactured parts, for
standardization and interchangeability of parts it is essential to
have low variability.

Therefore, we will consider several measures of dispersion, but

there are only two that are used much; the range and the standard
deviation.

Range
(c)

The range is the difference between the largest value and the
smallest value.
R = xn – x1
 Quantitative Techniques for Management Applications

304
For the example on leaving for office times, the range was 10

S
Notes
Activity
minutes (8.35 - 8.25).
What is range?
___________________

___________________
The problem with using range is that it considers the extreme
values only and does not use all the data in the sample. It is
___________________
therefore less reliable than some other measures of dispersion and

E
___________________ varies too much from sample to sample to be of effective use. It is
also very sensitive to the size of the sample, it usually increases
___________________
with the increase in the sample size, although not proportionately.
___________________
In spite of these shortcomings, there are special situations where

UP
___________________
the range is useful. When the sample is from a 'normal' universe
___________________ with a small sample size, the quantity is nearly as reliable as the
more laboriously calculated standard deviation. There are also
___________________
certain types of data and certain purposes for which the use of
___________________ range is appropriate, e.g., the range of temperature in Delhi.

Mean Deviation
Also known as average deviation, mean deviation is the mean of
the absolute amounts by which the individual items deviate from
E-
the mean. The following procedure is usually applied:

1. Calculate the absolute deviation from the mean, removing any

negative signs.

2. Sum all the deviations.

3. Divide the sum of the deviations by the total number of items.

CC

Symbolically, these steps may be summarized as follows:

Ex
MD =
n
Item X Deviation from mean Absolute deviation
number from mean
1 10 10 10
2 20 20 20
3 30 30 30
4 -10 -10 10
5 -20 -20 20
(c)

6 -30 -30 30
7 -25 -25 25
8 25 25 25
Total 0 0 170
UNIT 23: Basic Tools of Data Analysis–II

Ex 170 305
MD = = = 21.25

S
n 8 Notes
Activity
Mean deviation is simple and easy to understand and unlike R, it Make a difference between
___________________
M.D and S. D.
is affected by the value of each item. But it is unreliable because it ___________________
varies from sample to sample taken from the same universe. Also

E
___________________
it is a biased estimator of the population. Therefore standard
deviation, discussed below, is the most often used measure of ___________________

population dispersion. ___________________

___________________
Standard Deviation

UP
___________________
The standard deviation of a sample SD is similar to the mean
deviation in that it considers the deviation of each X value from the ___________________
mean. However, instead of using the absolute values of the ___________________
deviations, it uses the squares of the deviations. These are
___________________
summed, divided by n, and the square root extracted.
The formula for standard deviation (SD or s as it is usually
represented)
ΣX 2
E-
SD( σ ) =
n
Variance is the square of SD and is represented by:
ΣX 2
Variance (σ ) =
n

The detailed calculation is shown below:

CC

Deviation from Square of

Item number X
mean Deviation
1 10 10 100
2 20 20 400
3 30 30 900
4 -10 -10 100
5 -20 -20 400
6 -30 -30 900
7 -25 -25 625
8 25 25 625
Total 0 0 4050
(c)

X2 4050
SD (σ ) = = = 22.5
n 8
The concept of using sums of squares of deviations about the
arithmetic mean of a distribution is very important and we would
use it extensively in the units that follow.
 Quantitative Techniques for Management Applications

306
Coefficient of Variance

S
Notes

___________________ To get an indication of the variation that is related to the mean, we

divide the standard deviation by the mean to get the coefficient of
___________________ variance. This enables us to compare two groups, which have
___________________ different standard deviations and means more easily.

E
___________________ σ
Coefficient of Variance = × 100
μ
___________________

___________________ Skewness

UP
___________________
Skewness may be defined as the lack of symmetry or degree of
___________________ distortion from symmetry exhibited by a probability distribution.
___________________
Any measure of skewness indicates the difference between the
manner in which the items are distributed in a particular
___________________ distribution compared to a normal distribution.

The most useful measure of skewness is the Karl Pearson's

coefficient of skewness.

Karl Pearson’s Coefficient of Skewness

E-
Mean–Mode
Coefficient of Skewness =
S tan ndard Deviation

When the mode is not clear or where there are two or three modes,
the following formula is used:
3(Mean–Mode)
Coefficient of Skewness =
CC

S tan dard Deviation

Check Your Progress

Fill in the blanks:
1. ………………mean the ranges for which we classified the
number of items.
2. An …………..can be defined as a central value around
which other values of series tend to cluster.
3. ……………is defined as the reciprocal of the arithmetic
mean of the reciprocals.
(c)

4. ……………is that value which has the maximum

frequency (i.e. occurs most often) in a given set of
values.
Contd….
UNIT 23: Basic Tools of Data Analysis–II

307
5. ____________ is the value of that item in a series which

S
Notes
divides the series into two equal parts, one part
consisting of all values less and the other all value ___________________
greater than it. ___________________

6. Skewness may be defined as the lack of ____________ or

E
___________________
degree of __________ from symmetry exhibited by a
___________________
probability distribution.
___________________

Summary ___________________

UP
___________________
In this unit, we have studied commercial averages which include
the study of moving average, progressive average, quadratic ___________________

average. Further, we have focussed on 3M’s of statistics i.e. mean, ___________________

median and mode. Finally, we have considered mean, deviation ___________________
and standard deviation.

Lesson End Activity

E-
Do a comparative study amongst 3M’s of Statistics.

Keywords
Commercial Averages: Commercial averages are the applications
of averages in commercial situations.
Standard Deviation: The standard deviation of a sample SD is
CC

similar to the mean deviation in that it considers the deviation of

each X value from the mean.
Skewness: Skewness may be defined as the lack of symmetry or
degree of distortion from symmetry exhibited by a probability
distribution.

Questions for Discussion

1. Make a comparison between mean, mode and median.
2. What are the key methods of measuring dispersion?
(c)

3. Write a note on the following:

(a) Standard deviation
(b) Coefficient of variation
(c) Skewness
 Quantitative Techniques for Management Applications

308
Further Readings

S
Notes

___________________ Books
___________________ R S Bhardwaj, Mathematics for Economics and Business, Excel
___________________
Books, New Delhi, 2005

E
___________________ D C Sanchethi and V K Kapoor, Business Mathematics

___________________ Sivayya and Sathya Rao, An Introduction to Business Mathematics

___________________
Web Readings

UP
___________________
http://www.math.ucla.edu
___________________
www.textbooksonline.tn.nic.in
___________________

___________________
www.mathbusiness.com
E-
CC
(c)
UNIT 24: Forecasting

Unit 24
309

S
Notes

Forecasting
___________________

___________________

E
___________________
Objectives
___________________
After completion of this unit, the students will be aware of the following
topics: ___________________

\ The mathematical model of forecasting ___________________

UP
\ Correlation analysis ___________________
\ How regression analysis works?
___________________
\ Multiple regression analysis and their applications in business
___________________

___________________
Introduction
The future has always held a great fascination for mankind.
Perhaps this is biologically determined. Man and the higher apes
seem to have brains that are equipped to engage in actions for
E-
which a future reward is anticipated. In extreme situation reward
is anticipated not in this life but in the next life.
There are two methodologies to anticipate future. They are called
qualitative and quantitative. But both start with the same
premise, that an understanding of the future is predicted on an
understanding of the past and present environment. In this unit,
CC

we will mainly deal with quantitative methods. We will also

distinguish between forecast and prediction. We use the word
forecast when some logical method is used.
The quantitative decision maker always considers himself or
herself accountable for a forecast—within reason. Let us look at
the conceptual model first and then the mathematical model and
algorithms in turn which are used for making forecast.

The Conceptual Model

The qualitative school has generated many philosophical, religious
(c)

or political conceptual models according to which the ideology and

dogma is structured and forecasts prepared. Quantitative decision
making, defined here as anything that is not based on underlying
belief, offers three conceptual models. They are quite quantitative
 Quantitative Techniques for Management Applications

310 to highly technical. They are guesstimate, fundamental and

S
Notes
Activity technical models.
What is Quantitative decision
___________________
In the guesstimate conceptual model the forecast is based on
making?
___________________ expert opinion. It is almost like qualitative decision making except
___________________ that the bias of many is pooled. This method of forecast basically

E
___________________
revolves around Delphi Method.

___________________ This conceptual model for forecasts should not be used when ample

___________________
data bases are available. It is also known as option methodology.
The Delphi method consists of a panel of experts and a series of

UP
___________________
rounds during which forecasts are made via questionnaire.
___________________
Whether expertise or ignorance is pooled in each round, the result
___________________ is the same: a forecast is born. But in the absence of sufficient
___________________ data, it may be preferable to develop heuristics first rather than to
rely initially on guesstimates.

The second conceptual model stresses the fundamentals that

impinge upon the environment at any given time. In this case the
E-
forecaster tries to ascertain the functional relationships among
variables defining the environment. In addition, attention is paid
to changes in the magnitude of the variables that make up the
environment. This conceptual model is superior because it is based
on logical considerations and not merely on expert opinion.

The reason why not all forecasters wholeheartedly embrace the

CC

fundamental conceptual model is that it takes a pretty good mind

to understand the variables and their interrelationships that
represent the environment. It takes constant study, constant
learning, constant testing and then the intellectual ability to
synthesize it all. To cite an example, it does not take much to come
up with a fundamental conceptual model to forecast a nation’s
economic activity. We know that Gross National Product (GNP) is
a function of consumption (C), investment (I), government
spending (G) and net exports (E). In equation form it appears as
GNP=C+I+G+E. Now each variable, i.e., consumption, investment,
(c)

etc., can be carefully quantified. For example, do we consume more

goods or services? More hard or soft goods? and so on. Beautiful
econometric models have been generated on the basis of this
conceptual model. Beautiful forecasts have also been presented.
UNIT 24: Forecasting

311
The third conceptual model was called technical. It is used by

S
Notes
forecasters who call themselves technocrats. Whenever, a Activity
What is the role being played
predetermined parameter that the technocrats follow reaches a ___________________
by mathematical models?
certain magnitude, they forecast a change in the environment ___________________

irrespective of the behaviour of other variables. Sometimes, this

E
___________________
model gives accurate results and some times not. ___________________

Check Your Progress ___________________

Fill in the blanks: ___________________

UP
___________________
1. In the guesstimate conceptual model the forecast is
based on ___________. ___________________

2. The qualitative school has generated many ___________________

philosophical, religious or __________ conceptual models ___________________

The Mathematical Model

The mathematical models play a very important role in
E-
forecasting. The quantitative analysis that underlies a forecast is
based on the type of conceptual model that has been chosen. The
guesstimate and technical models typically result in mathematical
models that are less rigorous than the fundamental model,
although the decision tools of the former may be applied in the case
of the latter as well. Again in this unit only those decision tools are
CC

discussed that are considered efficient and wisely used. The

mathematical model that is considered here should be used in
conjunction with the fundamental conceptual model.

The logic used in mathematical model is twofold. First, it is based

on the idea that the past and present environments may be used to
extrapolate the future, the forecasted environment. Second, the
environment itself comes about because of a functional relationship
that exists between the variable whose value is to be forecasted
and one or more other variables that determine the forecasted
variable’s magnitude. The environment can be an economy, the
(c)

market for a product, the productivity of a shift of assembly line

workers, the track performance of 1500 meter runners and so on. It
is clear that in forecasting there are always two types of variables:
the one that is being forecasted and one or more from which the
forecast is made. The first one is known as the dependent variable,
 Quantitative Techniques for Management Applications

312 the latter as the independent variable(s). The functional

S
Notes
Activity relationship between the two can be visualized within a system of
What is Simple Forecasting
___________________ coordinates where the dependent variable is shown on the y and
System?
___________________ independent variable(s) on the x-axis. Since both types of variables
have usually positive values, the entire environment (past, present
___________________

E
and future) is shown in the first quadrant. Those variables that
___________________
may affect the dependent variable’s magnitude but which are not
___________________ considered in the decision space of the forecast either because of
___________________ oversight or their effect is deemed negligible, are known as

UP
___________________ intervening variables.

___________________
Check Your Progress
___________________ Fill in the blanks:
___________________
1. The quantitative analysis that underlies a forecast is
based on the type of ___________ model that has been
chosen.
2. The functional relationship can be visualized within a
E-
system of coordinates where the dependent variable is
shown on the _____ and independent variable(s) on the
________ axis.

Simple Forecasting System

Such a system may be called a simple forecasting system. If there
CC

is more than one independent variable, a multiple forecasting

results. Visualize each independent variable as representing a
dimension in the decision space. For the case y=f (x1, x2), the space
may still be shown on a plane (a two-dimensional piece of paper). If
there are more than two independent variables then imagine a
different dimension for each variable—all starting at the origin
(O).
Now the question arises as to how the dependent and independent
variables may be calculated mathematically so that a forecast can
be made. Let us stay with the simple system (only one independent
(c)

variable x) for example purposes and argue that we can forecast

gross national product on the basis of consumption alone, that is,
GNP=f (Consumption). In developed economy we would not be too
far off the mark. Past and present GNP and corresponding
consumption data may then be obtained and plotted. The resulting
cluster of coordinates is known as a plot or scattergram.
UNIT 24: Forecasting

313
Now suppose that someone asked you to forecast GNP, given a

S
Notes
certain consumption value. Such a value may be obtained by
polling a sample of consumers about the amounts that they are ___________________
planning to spend. Then calculate the sample mean, construct a ___________________
confidence interval, and then an average line or curve may be

E
___________________
drawn through the plot and the y-value obtained by extrapolation.
The immediately apparent problem is the mathematically proper ___________________

selection of the line or curve, because there are many possible ways ___________________
of drawing such an average, as shown and hence many possible ___________________

UP
forecasted GNP values. There are two basic methods to solve this
___________________
problem. They are known as regression and smoothing. Each one
has spawned a number of offspring. In this book only those are ___________________
discussed that minimize the possibility of injecting the decision ___________________
maker’s bias into the forecast. Furthermore, the decision tools that
___________________
are illustrated represent the strongest mathematical link in the
chain.
Both methods use the arithmetic mean which represents the
forecasted value. Fitting an average line by freehand process,
E-
points A and C fall above while points B and D fall below this
average line. The distances between them and the line are shown
as a, b, g, d and the points a, b, c, d determine the line with its
familiar equation yc =a + dx1 where yc is any estimated (forecasted)
y-value given a certain x-value, a is the y-intercept and b the slope
of the line. Obviously the best and tightest fit is obtained when the
CC

sum of the distances equal 0, since the positive distances (above

the line) and negative distances (below the line) would be offset
against each other. Mathematically the tightest fit is obtained by
the method of least squares,
a2 + b2 + g2 + d2 = |Minimum|
This conditions holds when the normal equations are used for
calculating the a and b values, or, for the straight line where all
values were previously defined. The method is known as regression
analysis and was developed by Francis Galton. You may read a
book on statistics for deeper understanding of regression and
(c)

correlation analysis.
It should be noted that the known environment extends between
Points A and D. Given this environment and the line that has been
fitted to it, a forecast yc may be made given x as shown in Figure
24.1. However, we are now extrapolating into an unknown
 Quantitative Techniques for Management Applications

314 environment, that is, its real x and y values are not known.

S
Notes Therefore, we must make a crucial assumption, namely, for our
___________________ forecasted yc value to be correct, no material alternations in the
___________________ functional relationship y=f(x) must have taken place. If, for
example, a much larger or smaller real value corresponds to each
___________________

E
real x value in this unknown environment or if a new important
___________________ variable has entered the decision space, the forecasted yc value is
___________________ probably false.
___________________

UP
___________________

___________________

___________________

___________________
E-
CC

Figure 24.1: Fitting a line yc = a+bx, to four points A, B, C, D

As we will see shortly, fitting lines and curves to given data sets is
simple. In fact the computer usually does it for us. Technocrats do
this very well. But the good forecaster knows his environment. The
skill lies in understanding the interrelationships in a multi-
variable decision space and to know when a change can be
expected in the functional relationships or when new variables
(henceforth intervening variables) must be considered an integral
part of that decision space. It is in this aspect of forecasting that a
(c)

solid fundamental conceptual model pays its dividends. A good

forecaster never loses sight of the fundamentals. The master
forecaster already knows the fundamentals of the unknown
environment. This is a skill that cannot be taught. It can only
“happen” to the individual who is willing to completely immerse
himself in a veritable flood of data and information blocks that
UNIT 24: Forecasting

may in any conceivable way have some bearing on the forecast. If 315

S
he ever comes up again, it has happened: one of the few master Notes
forecasters has joined the ranks. ___________________
A distinction is made between linear and curvilinear regression ___________________
analysis. Linear analysis fit straight lines to data sets. Curvilinear

E
___________________
or nonlinear analysis does the same with curves. Furthermore
there is a simple regression analysis with only one independent ___________________

variable, i.e.; y=f (x), and multiple regression with more than one ___________________
independent variables or y=f (x1, x2,..., xn). A special type of ___________________

UP
regression analysis uses time as the independent variable. Sales
___________________
forecasts are examples where Sales =f (Time). This type of
regression analysis is known as time series analysis. Finally, ___________________

before turning to the algorithms, a word may be added about ___________________

smoothing, a major forecasting method. As the name implies ___________________
observed values are smoothed and a weighted average is obtained
which represents the behaviour of the variable under
consideration. It is to be noted that the smoothing method is
mathematically much less rigorous than regression analysis. Also,
E-
the method is limited in its scope of application.

Check Your Progress

Fill in the blanks:
1. If there are more than one independent variable, a
_________forecasting results.
CC

2. ___________ each independent variable as representing

a dimension in the decision space.

Algorithms and Applications

Each of the two forecasting methods—regression and smoothing—
have their distinct algorithms. They will be discussed in turn. In
order to set the mental stage for this discussion, it is helpful to
rethink the approach taken in the particular type of forecasting
analysis that is examined here. It is not a forecast out of the blue.
Rather it is based on functional relationships among variables for
(c)

which there is a stipulated logic. In other words there is a

fundamental conceptual model upon which the quantitative
analysis is based. Furthermore it is assumed that the past and
present environments, that is, the joint behaviour of the dependent
and independent variables, are indicators of the future
environment. Mathematically speaking, the forecast is nothing but
 Quantitative Techniques for Management Applications

316
an extrapolation of the past and present. As previously pointed

S
Notes out, if no fundamental changes take place, with respect to the
___________________ magnitude of the functional relationships among the variables and
no new variables enter the decision space, this logic has been found
___________________
to work well when applied. It has stood the test of time. But the
___________________ above assumption also means that the direction of the forecast,

E
___________________ that is, an increase or decrease in the magnitude of the forecasted
behaviour of the variable, must be clearly visible and
___________________
mathematically substantiated in the past and present
___________________ environments.

UP
___________________
The Regression Algorithm
___________________
It may be recalled that, depending upon the forecasting problem,
___________________
regression analysis may take several forms. There is linear and
___________________ non-linear regression. There is simple and multiple analyses. And
there is time series analysis. For each, however, the same
algorithm or solution methodology may be used. This algorithm
has following steps.
E-
(1) Prepare a plot.

(2) Fit a line or curve to the plot and define either mathematically
by the method of test square.

(3) Test the significance of the slope. Sometimes people skip this
test.
(4) Construct a confidence interval for the forecasted yc value.
CC

(5) Estimate the quantitative effect of the independent variable(s)

x on the behaviour of the dependent variable y. This is known
as correlation analysis.

(6) Test the significance of the correlation. This, too, is sometimes

ignored.
After you have done this, what have you got? You got yourself a
forecast and you may be 95% confident (remember that in
forecasting it is always 95%) that your forecast will be on the mark
when, that mark ultimately becomes known assuming that the
logic of your conceptual model is sound. This is a big assumption,
(c)

because anybody can come up with a functional relationship y=f(x);

which may not necessarily be sound.
Let us proceed in our examination of the decision tools from the
simple to the more complex ones and explain each step of the
algorithm in detail as we go along. We begin with simple linear
UNIT 24: Forecasting

regression and correlation analysis, yc=f (x), then switch to linear 317

S
and non-linear time series analysis, yc=f (x, time), which is only a Notes

special case as may be recalled, and finally look at multiple ___________________

regression and correlation analysis, yc=f(x1,x2,...,xk). ___________________

E
___________________
Check Your Progress
___________________
Fill in the blanks:
___________________
1. Depending upon the forecasting problem, regression
analysis may take _________ forms. ___________________

UP
2. There is _______ and __________ regression ___________________

___________________

Simple Linear Regression Analysis ___________________

A time honoured functional relationship exists between the ___________________

amounts spent on advertising and sales generated by these

amounts. It is a popular belief that this is a positive relationship in
the sense that each rupee spent on advertising generates so many
additional rupees (say, about 10 for consumers’ goods) in sales
E-
revenue. Indeed, the relationship appears to be linear within
meaningful ranges of the advertising budget. Thus, very little
money spent on advertising may have very little effect on sales or
none at all. Similarly, extremely large sums that are spent on
advertising will not generate that much more in sales revenue.
But, a meaningful advertising budget (meaningful in terms of
CC

market constraints, demand factors, customer income, etc.),

usually shows a linear relationship to sales revenue. In order to
perform regression analysis, data sets of between n=15 and n=25
should be used as a minimum sample size. After all, the past and
present environment is described and too small a sample of
observations will not do. For example purposes we will use only
n=5 so that the manual calculations do not detract from an
understanding of the procedures. On the job, regression analysis
should be performed on the computer exclusively once the
method has been understood. For easier referencing all
calculations for each step of the regression algorithm
(c)

are shown in one work sheet with the column number identified for
each step. Let us get to work on an assignment that has been just
received from the controller of our company. Given advertising
expenditures of ` 8 million for the next fiscal year, how much
 Quantitative Techniques for Management Applications

318 sales revenue can be expected? Is the question that the company

S
Notes wants an answer for?
___________________
Let us try to answer this question. First, the conceptual model:
___________________ what, if any, relationship exists between sales and advertising?
___________________ This has been taken care of. We know that sales =f (Advertising).

E
Secondly, how does this relationship look in the case of our
___________________
particular firm? OK. We need data. We get the data from our
___________________ accounting department. They tell us that in the past, five
___________________ advertising expenditures (budgets) resulted in certain

UP
corresponding sales figures. Remember that we use n=5 in order to
___________________
keep the calculations simple. Notice that they gave us a random
___________________
sample of observations that is deemed representative of the
___________________ environment. The data are recorded in ascending order (not a
___________________ requirement) in the work sheet below. Sales are the dependent
variable and advertising the independent variable. Already at this
point begin to think about the intervening variables that are likely
to be operative in this decision situation. We will have to come
back to them later on. And now let us activate the algorithm.
E-
Plotting the Data
The data, as they appear in columns 1 and 2 of the worksheet, are
placed into a system of co-ordinates. This can be done manually or
by computer. When making a forecast about a certain phase of
your operations, you usually have a pretty good idea of how the
result “ought” to look. This mental picture is the result of your
CC

collective experience with the operations. The plot allows you to

verify the mental picture with reality. You can see, for example, if
there is a positive, negative or zero relationship between the
variables. You can see whether the relationship is consistent
(small variance) when the data cluster is close or inconsistent
(large variance) when they are all over the quadrant. These
observations help you to shape the conceptual model.
Also, you can see if by freehand method you may want to fit a line
or curve to the data as previously mentioned. Furthermore you can
see whether linearity or non-linearity governs the entire data set
(c)

or if there are combinations. These observations help you to decide

on the Mathematical model for your forecast. Obviously your choice
of a linear versus non-linear analysis has great impact on the
forecasted value and the implications of this choice will be
discussed in detail when performing the non-linear analysis later
on.
UNIT 24: Forecasting

319
Table 24.1: Worksheet for a Simple Linear Regression Analysis

S
Notes

___________________

___________________

E
___________________

___________________

___________________

___________________

UP
___________________

___________________

___________________

___________________
E-
CC
(c)

A good forecaster will think about the plot and inspect it again and
again for a considerable period of time before fixing in his mind the
 Quantitative Techniques for Management Applications

320 conceptual and mathematical models. Once the models are

S
Notes determined, the rest of the quantitative analysis is routine. The
___________________ work can be turned over to the computer. But relating a mental
picture to reality and vice versa goes a long way in making a valid
___________________
forecast. The plot helps in this endeavour.
___________________

E
Our plot is shown in Figure 24.2. Disregard for the moment
___________________
everything except the connected original data points shown as
___________________ circles. These data are taken from the worksheet. You
___________________ can readily see that if you had to fit an average to this data set by

UP
freehand method, you would use a straight line. The line that
___________________
represents the best fit is calculated next in Step 2 of the algorithm.
___________________

___________________

___________________
E-
CC

Figure 24.2: Plot of Original Values and Line of Best Fit

Fitting the Straight Line

The mathematically best fit of the line yc=a+bx to a data set is
obtained by the method of least squares as previously discussed.
The calculations for the normal equations
Σ y = na+ b Σ x
2
Σxy = a Σ x +b Σ x

are shown in the worksheet resulting in

(c)

245 = 5a + 25b

1327 = 25a + 136.98b

which now may be solved simultaneously by solving for b and

substituting in the first equation in order to find a. Thus,
UNIT 24: Forecasting

321
[245 = 5a + 25b ] × 5

S
Notes

⇒ 1225 = 25a + 125b ___________________

1327 = 25a + 136.98b ___________________

E
___________________
102 = 11.98b
___________________
b = 8.5142
___________________
Therefore, the best fitting line has been defined as ___________________
yc=6.429 + 8.5142x.

UP
___________________
Rather than to solve a and b algebraically, the normal equations ___________________
may be solved for a and b. After simplification
___________________
nΣxy - ΣxΣy
b= ___________________
nΣx 2 - ( Σx) 2

is obtained. Using these direct formulae which are more efficient

(5)(1327) - (25)(245)
b=
E-
2
(5)(136.98) - (25)

510
=
59.9

= 8.5142

and
CC

245 25
a= - (8.5142) = 6.429
5 5
The line yc=6.429 + 8.5142x is called the forecasting equation. It is
the decision tool that allows us to answer the controller’s question.
You remember that she stipulated an advertising budget of ` 8
million or x=8.

We can now estimate sales revenue by

yc = 6.429+(8.5142) (8)
(c)

= ` 74.5426 million

You know that the straight line is defined by two points. Given the
forecasting equation, pick any two (simple) x-values. Let us say
 Quantitative Techniques for Management Applications

322
that we use x=0 because this is the y-intercept and x =5.5. The

S
Notes
estimated y-values are then found by
___________________
yc = 6.429+ (8.5142) (0)
___________________
= 6.429
___________________

E
___________________
and

___________________ yc = 6.429 + (8.5142) (5.5)

___________________ = 53.2571

UP
___________________ These values are shown in Figure 24.2 as squares and the
___________________ regression line has been constructed. As you read up to the line at
x=8; you find the forecasted yc –value. Note the managerial
___________________
meaning of the y-intercept a and the slope b. At zero advertising
___________________
expenditures, sales amount to a= ` 6.429 million. This tells you
that there is not exactly a perfect relationship between sales and
advertising. If there were, you would see zero sales. You are still
thinking about the intervening variables? Always do. Now, the
slope b= ` 8.5142 million tells you that as the controller authorizes
E-
the advertising budget, each Re. 1 million will result in ` 8.5142
million in sales. You realize, of course from the preceding units,
that since only sample data are used, this value must be seen as
the midpoint of a confidence interval. So don’t call the controller
and say ` 1 million advertising results exactly in ` 8.5142 million
in sales. This is stochastic decision making after all. Find the
CC

standard error of the slope and look up the proper t - value and put
everything into the 95% confidence interval for the regression
coefficient b (note the new term) which takes the form

b ± t 0.025 σb
While you are at it, you suddenly remember (from the preceding
unit) that the forecasted sales value yc= ` 74.5426 million should be
communicated to the controller in confidence interval form as well.
We are still talking sample statistics and always will be. We never
know the population regression line Y=a+bX. All we know is the
sample regression line yc=a+bx. So, better hold that call and let us
(c)

figure out σb first. Needless to say that you could also calculate σa—
if the urge ever struck you. Should it strike you? Look at it this
way. At the y-intercept (a), x=0. But you postulated y=f (x)! If you
want to set x=0, you don’t have much of a relationship. Do you?
You want to study the effect x has on y, and therefore, you are
UNIT 24: Forecasting

interested in the slope b. Leave σa to the statisticians. They have 323

S
the most peculiar urges at times. Notes
Activity
Discuss significance test of
___________________
The Significance Test of the Regression Coefficient b the regression coefficient.
___________________
So you are interested in the slope b. The question is at what angle

E
___________________
to the x-axis a slope becomes significant. Only a significant slope
means that y=f(x) is real in the stochastic decision making sense. If ___________________
the slope is significant case there is a forecasting tool. In the ___________________
insignificant case there is nothing except a waste paper basket for
___________________
the study or smoothing.

UP
___________________
Regression analysis is based on the assumption that the y-variable
___________________
is normally distributed. Figure 24.3 shows a pictorial presentation.
Then, it will be recalled from the preceding unit, the proper ___________________
significance test for b is the one-sample t-test. ___________________

The algorithm for that test is as follows:

Step 1: Ho : b = 0

Ho : b > 0
E-
or

B≠ 0
It is left to the discretion of the forecaster whether to use a one or
two-sided alternative hypothesis. In this particular case the plot
shows clearly a positive slope. Therefore to test b < 0 does not
CC

make much sense. This test, however, is included in b ≠ 0. So why

use it? The answer is that it is always easy to reject Ho when using
a one-tailed test. Remember the mathematically expected values
(MEV) at P .05 for the one and two-tailed tests? They are 1.64 and
1.96, respectively. You can see that a smaller experimentally
obtained value (EOV) leads to the rejection of Ho. Take, for
example, EOV=1. 75. Then with a one-tailed test you reject. But
not with the two-tailed test. Rejecting Ho in the regression problem
means that the regression coefficient b is significant which means
in turn that the regression equation may be used as a valid
(c)

forecasting tool. A conservative forecaster may argue that the

selection of such a tool should be as severely constrained as
possible. Hence, the two-sided alternative hypothesis.
 Quantitative Techniques for Management Applications

324
Step 2: P.05

S
Notes

___________________ The 95% confidence level is selected as standard operating

procedure in forecasting studies involving regression analysis for
___________________
reasons previously stated.
___________________

E
b-0
___________________ Step 3: t =
σb
___________________

___________________ The one-sample mean problem formula for the t-test may be
recalled.

UP
___________________
It reads
___________________

___________________
x-μ
t=
x
___________________
Remember that each point on the regression line with slope b
represents a mean. Thus, we are talking, in the regression case,
simply about a “many-sample” mean problem where each data
consists of two values, x and y. So you can see the similarities in
E-
the two t configurations: b takes the place of x, o the place of
because we want to test whether the population regression
coefficient equals zero (this is the null hypothesis in all regression
problems) and finally the standard error of the regression
coefficient b, σb is substituted for the standard error of the mean, σ
You recall that σ is unknown and must be estimated by
CC

s
σx =
n

Similarly, σb must be estimated, but this estimation is somewhat

more complex because of the broader scope of the problem. It is
Syx
2
σ b = Z(x-x)2

where σ2yx is the standard error of the estimated regression

equation of the y values on x. This standard error σyx is defined by

Σ (y - y c )
2
2
S =
(c)

yx
n-m

Where n-m are the degrees of freedom with m the number of

regression coefficients. In the case of the straight line m=2 because
there are two quantities a and b. The calculations for Step 3 of the
algorithm may now be performed as shown in Table 24.1 columns 5
UNIT 24: Forecasting

through 7. Note the distinction between the observed y value 325

S
(column 1) and the estimated yc value (column 5). The latter is Notes

obtained by solving the regression equation yc=6.429+8.5142x for ___________________

every observed x value (2.8, 4.3, 5.0, 5.5, 7.5) as shown in column 2. ___________________
In other words every estimated yc value is obtained for every

E
___________________
observed x value. Then, sum the squared difference between y and
___________________
yc as shown in Table 24.1. The SS of the x values (column 7) you
calculated in the preceding unit. Note that as shown in column 2, ___________________
the t test may now be performed as follows, ___________________

UP
2 51.5524 ___________________
Syx =
5-2
___________________
= 17.1841 ___________________

and ___________________
17.1841
11.98

=1.1977
E-
If still interested, you may now construct the confidence interval
for the slope b. Finally

8.5142 - 0
t=
1.1977

= 7.11σ
CC

IV EOV>MEV or 7.11 > 2.353 (one-tailed)

or 7.11 > 3.182 (two-tailed)

∴ The slope is not zero.

∴ The regression equation may be used as a valid forecasting tool.

The Confidence Interval for yc

We are now ready to calculate a proper stochastic answer to the

forecasting problem at hand. Remember that since sample data are
used, the answer always takes interval form. The 95% regression
(c)

interval is

yc ± Sd (t.0250)

yc ± Sd (t.0250)
 Quantitative Techniques for Management Applications

326
Where Sd is the standard error for a certain yc value given its

S
Notes
corresponding x value.
___________________

___________________ This standard error is defined by

2
___________________ 1 (x g - x)
[ n + Σ (x - x) ] S

E
2 2
Sd = 2 yx
___________________

___________________ All values are known. Note the xg value in the numerator which is

___________________
the given x value corresponding to the yc value. In other words for

UP
each yc, a new interval must be constructed. As shown by Figure
___________________
24.3 each point on the regression line represents the mean of a
___________________
normally distributed sample of y values. The upper and lower
___________________ confidence limits of these samples are not parallel to the regression
___________________ line but are curvilinear as shown in somewhat exaggerated form.
E-
CC

Figure 24.3: Normal Distribution of y-values

Now, our controller wants to know the forecasted sales revenue (yc)
given an advertising budget of ` 8 million (x=8). We calculated yc =
` 74.5426. Calculating first the standard error
2
2 1 (8 - 5)
S2 =[ + ] 17.1841
5 11.99

=16.3464

the interval is
74.542 ± ( 16.3464) (3.182)
(c)

74.542 ± 12.8650

or

(61.677, 87.407)
UNIT 24: Forecasting

327
Remember that t is distributed with (n-m) or 3 degrees of freedom

S
Activity
Notes
and that, since an interval is involved with upper and lower limits, Discuss the parametric case
the two-tailed MEV must be employed. ___________________
of correlation analysis.
___________________
And now our final answer to the company: we tell the controller
that we are 95% certain that with an advertising budget of ` 8

E
___________________
million and assuming no major changes in the market ___________________
environment, between ` 61.677 and ` 87.407 million in sales
___________________
revenue may be expected.
___________________
But the sales revenue may be the result of some other factor than

UP
advertising. ___________________

___________________
Check Your Progress
___________________
Fill in the blanks:
___________________
1. It is a popular belief that this is a ___________
relationship in the sense that each rupee spent on
advertising generates so many additional rupees (say,
about 10 for consumers’ goods) in sales revenue.
E-
2. Each point on the regression line with slope b represents
a _________.

Correlation Analysis (The Parametric Case)

A correlation analysis which measures the closeness of fit of the
CC

regressions line while assuming that both x and y are normally

distributed (bivariate normal distribution) is necessary to set at
rest the above doubt. Hence a parametric analysis is performed.
Correlation analysis may also be seen as a measure of mutuality of
x and y. Indeed, it is the more prevalent approach and computer
programs are usually based on it. Of course, the results are the
same. Let us start with closeness of fit because we have most of the
calculations completed and the reasoning process of fitting a line is
still fresh in our minds. But first a general word about this new
decision tool.
The degree of correlation between normally distributed dependent
(c)

and independent variables is signified by the correlation coefficient

r. The symbol r was used in order to honour Francis Galton’s work.
You recall that he gave us regression analysis and coined the term
regression. Perhaps it would have made more sense to call the
regression coefficient rather than the correlation coefficient r. But
 Quantitative Techniques for Management Applications

328 the slope of the line was already labelled (b is generally known as

S
Notes the regression coefficient although there is a as well), so the
___________________ correlation coefficient r it had to be. The coefficient r is a pure
number. It is constrained by ±1 and defined by
___________________
2
___________________ Sxy

E
r = 1- 2
___________________ Sy

___________________
Where S2y is the variance of the y values. But since these y-values
___________________
are part of a regression problem, n – m degrees of freedom apply.

UP
___________________ Therefore, in the case of the straight line, m=2 previously
___________________ discussed and as shown in Table 24.1 columns 8 and 9. Then
2
___________________ z Σ (y i - y)
Sy =
___________________ n-m
2
Σ (y i - y)
=
n-2
920
=
3
E-
= 306.6667

As pointed out, the correlation coefficient r is a pure number. Its

sign is positive in this case because the slope of the regression line
is positive. If b were negative, r would be negative as well. This
fact is mentioned because it is not readily apparent since the
radical has both the positive and negative sign.
CC

Various types of correlation and their respective plots are shown in

figure 24.4
(c)

Figure 24.4
UNIT 24: Forecasting

Since perfect positive correlation is r=+1.00, but r is a pure 329

S
number, we cannot specify an exact quantitative meaning of the Notes

value that we obtained, r = + 0 .9716. We can only say in general ___________________

terms that our findings show a “high” positive correlation between
___________________
sales and advertising. In order to establish a specific quantitative

E
___________________
relationship, we must calculate r2=0.9440. This is the coefficient of
determination which indicates the percentage of the variation in ___________________
the y variable that is caused by the x variable. ___________________

For purposes let us say rather loosely—the specifics are discussed ___________________

UP
in the section about the significance of r —that 94.40% of the
___________________
increases or decreases in the sales revenue are caused by
___________________
advertising expenditures and only 5.6% are caused by other
intervening variables. These findings should pacify our controller, ___________________
because our product sales seem to be very sensitive to advertising ___________________
expenditures. More is to be said about the significance of
correlation, let us hold this judgement for a while.
First, though, let us take a look at the other method, mutuality of x
and y of correlation analysis. It was mentioned previously that
E-
parametric correlation analysis underlies the assumption of
bivariate normal distribution. If this is the case, the distinction
between independent and dependent variable may be dropped. We
may perform a regression of y on x values—as we did in the
forecasting study—or regress x on y values. In the first case we
obtain our familiar equation yc=a+bx. In the second case we
CC

interchange the variable titles for the given data sets in columns 1
and 2 and obtain xc=a'+b' y. You may do this right now and plot the
two regression lines. Obviously, with perfect correlation the two
lines would be superimposed. Less than perfect correlation results
in two interesting lines at ever wider angles as the correlation
decreases. Obviously, the position of the lines are caused by the
slopes b and b'. And by now you probably sense already something
interesting, namely, that there is a direct relationship between the
correlation coefficient r and the two regression coefficient b and b'.
Indeed, the correlation coefficient is the geometric mean of the
regression coefficients, or r = bb1
(c)

Normally, this book does not discuss the derivation of the decision
tools. But in this case you can get a little flavour of that too
because a knowledge of higher mathematics is not necessary. You
recall that we solved the normal equations for b yielding
 Quantitative Techniques for Management Applications

330
nΣ xy - ( Σx)( Σy)

S
Notes b=
n Σ x 2 - ( Σ x) 2
___________________

___________________
This was simplified and in a form that uses totals rather than
deviations. The matter was discussed in the preceding unit,
___________________

E
formula for the variance. Remember? Now let us take one step
___________________ back and show the formulae for b and b' in deviation form.
___________________ Σ ( x - x) (y - y)
b=
___________________ Σ ( x - x) 2

UP
___________________ and
___________________ 1
Σ (y - y) (x - x)
b =
___________________ Σ ( y - y) 2
___________________ then
[Σ(x - x) (y - y)]2
bb =
Σ( x - x) Σ( y - y)
2 2

and
E-
Σ(x - x) (y - y)
bb =
Σ( x - x) 2 Σ( y - y) 2

or in simplified “totals” form

nΣ xy - ( Σx) ( Σy)
r=
[nΣx 2 - ( Σx)2 ][nΣy 2 - ( Σy)2 ]
CC

This formula is known as the Product-Moment Formula or

Spearman- Brown Formula. It is identified as such in the computer
libraries. Solving for our sales and advertising example, using the
calculations in Table 24.1. We obtain

(5)(1327) - (25)(245)
r=
[(5)(136.98) - (25) 2 ] [(5)(12925) - (245) 2 ]
510
=
524.919
= 0.9716
(c)

which is the same value as obtained previously.

A concluding comment about parametric correlation analysis may

sharpen further our understanding of the subject. Remember that,
the idea behind correlation analysis is to specify the variation in
UNIT 24: Forecasting

the dependent variable that is caused by the independent variable 331

S
or variables in the multiple correlation case as discussed later. We Notes

may show this idea in basic equation form like this - ___________________

Total Variation = Unexplained + Explained ___________________

E
___________________

___________________

___________________
Variation Variation
___________________

UP
Where the explained variation is the portion that has been
___________________
explained by the regression of y on x, or, 94.40% in our example
and the unexplained variation, equal to 5.6%, which is due to the ___________________

influence of intervening variables and statistical error. In symbolic ___________________

form this equation may be written as
___________________
Σ (y - y) 2 = Σ (y - y) 2 + Σ (y c
- y)
2

The coefficient of determination is the ratio

2 ExplainedVariation
E-
r =
TotalVariation

and perfect correlation would exist, r2=1.00, if the unexplained

variation equalled zero. Rewriting the symbolic form of the basic
equation with the degrees of freedom and previously introduced
symbols we obtain
CC

2 2 2
S y = S yx + S y(c)

Where S y2( c ) is the variance of the yc values and has not been
calculated, but it is known by substraction if S y2 (total variance)
2
and S yx (unexplained variance or standard error) are known. This
is indeed the case and r2 may be solved in the closeness of fit
approach.
2
Σ (y c - y)
2 n-m
r = 2
Σ (y - y)
n-m
(c)

2
S y(c)
= 2
Sy
2 2
S y - S yx
= 2
Sy
 Quantitative Techniques for Management Applications

332
and, as shown before,

S
Notes
Activity
2
Discuss the non parametric
___________________ 2
S yx
case of correlation analysis. r = 1- 2
___________________ Sy

___________________

E
Check Your Progress
___________________
Fill in the blanks:
___________________
1. Correlation analysis may also be seen as a measure of
___________________
___________.

UP
___________________
2. Symbol r was in correlation analysis to honour
___________________ ___________ ___________ move.
___________________

___________________ Correlation Analysis (The Non-parametric Case)

When the bivariate normal distribution cannot be assumed or the

observed data represent rankings, a non-parametric correlation
analysis must be performed. The reasoning that underlies this
E-
technique is the same as discussed in the previous unit. The
appropriate decision tool is known as the Spearman Rank Order
correlation coefficient.
2
6Σd
rs =1- 2
n(n - 1)
CC

Where rs stands for the Spearman r and d represents the

differences between ranks as already encountered in the previous
unit. In order to illustrate the application of this tool, let us use the
problem on supermarket stools. Ten cashiers were tested for
productivity increase after a stool had been installed at the
supermarket checkout. The data for “with stool” and “without
stool” conditions are ranked, the squared rank differences obtained
and summed.

The worksheet in Table 24.2 shows the computations.

Table 24.2: Worksheet
(c)

With Without Rank Rank

Cashier d d2
Stool Stool “With” “Without”
1 21 21 2.0 3.0 -1.0 1.00
2 24 23 7.0 0.0 0.5 0.25
3 28 27 10.0 10.0 0.0 0.00
Contd…
UNIT 24: Forecasting

4 22 24 4.5 8.0 -3.5 12.25 333

S
5 22 21 4.5 3.0 1.5 2.25 Notes
6 25 25 8.5 9.0 -0.5 0.25 ___________________
7 21 20 2.0 1.0 1.0 1.00
___________________
8 21 23 2.0 6.5 -4.5 20.25

E
9 23 21 6.0 3.0 3.0 9.00 ___________________

10 25 22 8.5 5.0 3.5 12.25 ___________________

Total 58.5
___________________

Therefore, ___________________

UP
2 6(58.5)
r = 1- ___________________
10(10 2 - 1)
___________________
351
= 1-
990 ___________________
= 0.6455
___________________
It may be recalled that when this problem was phrased as a non-
parametric dependent two-sample mean comparison test, the
Wilcoxon’s Test proved to be insignificant. What do you think is
the significance or rs in this case? When the introduction of a stool
E-
did not make a difference in productivity, would it not make sense
to forecast the “with stool” condition by the “without” condition?
With Stool Productivity =f (without stool productivity)? Of course it
would. Now, what does this mean concerning the significance of rs?
You’ll find out in the next section. Meanwhile ponder the problem
and note that =0.4167 which means that 41.67% of the variation in
CC

with-stool-productivity is “caused” by without-stool-productivity.

Significance Test of the Correlation Coefficients r and rs

Correlation analysis examines the functional relationship between
dependent and independent variables. The question arises after an
r or rs has been calculated whether the relationship that is
expressed by it happens to be meaningful or not. The question is:
what is meaningful? Meaningfulness may be defined in a
statistical sense and an operational sense. And so it is in any
correlation analysis. The decision maker has to look at significance
(meaningfulness) from two perspectives. Let us look at the easier
(c)

one—statistical significance first.

As in the case of the regression coefficient b, the regression
coefficient r must be tested for significance; for, if it turns out to be
insignificant, only a change relationship exists between the
dependent and independent variables. In those correlation analysis
 Quantitative Techniques for Management Applications

334 that assume a bivariate normal distribution (parametric

S
Notes correlation), the t test may be used as before in the case of b. For
___________________ the Spearman Rank Order Correlation (non-parametric case),
significance values developed by E. Olds must be used. Let us run
___________________
the test for the parametric example first. The steps in the
___________________

E
algorithm are as follows:
___________________
Step 1: HO : ρ = 0 (Rho, population correlation coefficient)
___________________
HA: ρ¹ 0
___________________

UP
___________________
or

___________________ ρ>O

___________________ Step 2: P.05

___________________ Step 3:

r
t=
1- r2
n-2
E-
0.9716
=
1 - 0.9440
5-2
0.9716
=
0.1366
= 7.11σ
CC

Note that this t-value is exactly equal to the t-value that was
obtained before when testing the significance of b. Minor rounding
errors may be expected.

EOV > MEV or 7.11> 2.353 (one-tailed)

or 7.11> 3.182 (two-tailed)

∴ The correlation coefficient is not zero.

∴ There is a statistically significant correlation between

advertising expenditures and sales revenues.
(c)

The direct relationship between r and b was noted. Therefore the

equality of the t value for both r and b in Step 3 of the algorithm
does not come as a complete surprise. Obviously if b is known to be
significant, r is also significant and vice versa. You notice that the
significance test for r is much easier to perform manually than the
UNIT 24: Forecasting

test for b. It may pay therefore to test the significance of r if you 335

S
want to find out the significance of b. Similarly, as seen in the Notes
multiple regression case many computer programs provide only ___________________
the significance for b. If you happen to be interested only in a
___________________
correlation analysis, run the regression equation to obtain b, test

E
its significance and you know whether the r in which you are ___________________

interested is significant or insignificant. ___________________

Let us now turn to the nonparametric correlation example and the ___________________
application of the Olds’ tables. Our experimentally obtained value ___________________

UP
was rs =0.6455. Remember that we have an answer outstanding
___________________
from you. Using the two-sided alternative hypothesis that was
used in the corresponding means comparison test, we find in the ___________________

table, MEV= 0.564 for n = 10. Since EOV>MEV we have obtained a ___________________
statistically significant correlation. Was that your outstanding ___________________
answer a few pages back? If so, you are getting the hang of it. And,
by the way, it makes good sense. If the means difference had been
significant, all we could expect was a chance relationship for With-
Stool-Productivity= f (Without-Stool-Productivity).
E-
Now let us turn to the much more difficult answer concerning the
operational significance of a correlation. To give a comprehensive
answer, we consider three types of correlations, all of which may be
statistically significant but which have vastly different operational
implications for the decision. The first one is known as causal
correlation. Here one variable causes the behaviour of the other. Of
CC

course this is hoped for status in any regression study, e.g.,

advertising causes sales, or, sales = f (advertising). Could the
casualty be reversed? Nothing says that it can’t give the
assumption of a bivariate normal distribution. Indeed, we saw that
in the Product Moment formula that it did not make any difference
which variable was dependent or independent. Then sales cause
advertising. Why not? It is the old chicken and egg story. But note
that you and the controller would approach the decision itself
(there is ` 8 million involved, after all) from a completely different
perspective.
The second type of correlation may come about because of co-
(c)

dependence. In this case the behaviour of the dependent and

independent variables is caused by an intervening variable or
variables. In the sales advertising study, sales = f (advertising)
may be statistically and operationally significant but only because
our customers have sufficient disposable income to buy. You can
 Quantitative Techniques for Management Applications

336 see that it takes much more for a good forecaster than to throw

S
Notes
Activity around a few formulae. It takes a sound conceptual model.
What is time series analysis?
___________________
The third type of correlation is due to coincidence. There is no
___________________ logical tie between the variables, but there may be a beautiful
___________________ statistical significance nevertheless. For example, the correlation

E
between the growth of a child and a plant such, correlations need
___________________
not be considered in decision making.
___________________

___________________
Check Your Progress

UP
Fill in the blanks:
___________________

___________________ 1. The first type of correlation is __________ correlation

___________________ 2. The second type of correlation is __________ correlation

___________________ 3. The third type of correlation is __________ correlation

Time Series Analysis

Time has strange, fascinating and little understood properties.
Virtually every process on earth is determined by a time variable.
E-
One of the most frequently encountered managerial decision
situations involving forecasting is to measure the effect that time
has on the sales of a product, the market price of a security, the
output of individuals, work shifts, companies, industries, societies
and so on. A fundamental conceptual model in all of these
situations is the product life cycle concept which goes through four
stages – introduction, growth, maturity and decline. Let us look at
CC

this concept in greater detail before we apply it.

(c)

Figure 24.5: Product Life-Cycle

UNIT 24: Forecasting

The sales performance of this product goes through the four 337

S
stages—introduction, growth, maturity and decline. Data have Notes

been plotted and regression lines fitted to each of the four ___________________

environments. Thus, when a sales forecast is made and the target ___________________
horizon falls within the same stage, the linear fit yields valid

E
___________________
results. If, however, the target horizon falls into a future stage, a
___________________
linear forecast may be erroneous. In this case a curve should be
___________________
fitted as shown. It is usually lightly speculative to select a
forecasting horizon that spans more than two stages. ___________________

UP
___________________
Another point of interest is the behaviour of the sales variable over
the short run. It fluctuates between a succession of peaks and ___________________

troughs. How do these come about? In order to answer this ___________________

question, the time series, must be decomposed. Then four ___________________
independent motors for this behaviour become visible. First, there
is a long-term or secular trend (T) which is primarily noticeable
within each stage of the cycle and over the entire cycle. Secondly,
cyclical variations (C) which are caused by an economy’s business
E-
cycles affect product sales. Such cycles, whose origins are little
understood, exist for all economies. Thirdly, the product’s sales
may be influenced by the seasonality (S) of the item, and finally,
there may be the irregular (I) effects of inclement such as weather,
strikes and so forth. In equation form the decomposed time series
appears as TS = T + C + S + I.
CC

This creates a complex situation in time series analysis. Each

factor must be quantified and its effect ascertained upon product
sales. Let us see how this is done. The long-term trend effect T is
reflected in the slope b of the regression equation. We already
know how b is calculated even though minor modifications of the
decision formulae will be encountered soon. The quantification of
the cyclical component C is beyond the scope of this book. However,
since business cycles always proceed from peak to trough to new
peak and so on, their positive and negative effects upon a product’s
sales cancel out in the long-run. Hence in managerial, as opposed
(c)

to economic, decision making, the sum effect of the business cycles

may be set equal to zero. This eliminates the C factor from the
equation. Seasonality, if present, is something that must be taken
into consideration because it is a product-inherent variable and
 Quantitative Techniques for Management Applications

338 therefore, it is under the immediate control of the decision maker.

S
Notes
Activity We will quantify the S component and keep it in the equation.
What is Target Horizons?
___________________
Finally, there are the irregular variations. Do we know in July
___________________ whether the weather will be sunny and mild during the four weeks
___________________ before Diwali? We don’t, but we know that if this happens, Diwali

E
___________________ sales will be severely impacted. Can we forecast such horrible

___________________
weather conditions? Not really. We cannot forecast them because
they cannot be quantified—a rather unpleasant characteristic they
___________________
share with all other type of irregular variations like strikes,

UP
___________________
earthquakes, power failure, etc. Yet, something strange usually
___________________ happens after such an irregular variation from “normal” has
___________________ occurred. Whatever people did not do because of it, like not buying
___________________ a product, they attempt to catch up with quickly. Therefore, the I
factor effect may also be assumed to cancel out over time and it
may be dropped from the equation which then appears to the
manager as TS = T + S.
E-
Check Your Progress
Fill in the blanks:
1. When a sales forecast is made and the target horizon
falls within the same stage, the ________ fit yields valid
results.
2. If, the target horizon falls into a future stage, a linear
CC

forecast may be _____________.

Linear Analysis
We will construct again the best fitting regression line by the
method of least squares. In order to illustrate the procedure, let us
use a data set from understand Case 24.1 at the end of the unit. It
involves the dividend payments per share of the Smart, a well-
known discount store chain, for the years 1990 through 1999.
Suppose that a potential investor would like to know the dividend
payment for 2001. The data are recorded in the work sheet (Figure
(c)

24.6) that appears below. First, however, turn your attention to

Figure 24.6 which shows the plot for this problem.
UNIT 24: Forecasting

339

S
Notes

___________________

___________________

E
___________________

___________________

___________________

___________________

UP
___________________

___________________
Figure 24.6: Plot of Dividend Values
___________________
Think for a moment about the qualitative nature of the time
___________________
variable. It is expressed in years in this case but could be quarters,
months, days, hours, minutes or any other time measurement unit.
How does it differ from advertising expenditures, the independent
variable that we examine in the preceding section? Is there a
E-
difference in the effect that a unit of each has on the dependent
variable, or, ` 1 million in one case and 1 year in the other? Time,
as you can readily see is constant. One year has the same effects as
any other. This is not true for advertising expenditures, especially
when you leave the linear environment and enter the non-linear
environments. Then there may be qualitative difference in the
sales impact as advertising expenditures are increased or
CC

decreased by unit.
Since time is constant in its effect, we may code the
variable rather than to use the actual years or other time units x
values. This code assigns a 1 to the first time period in the series
and continues in unit distances to the nth period. Do not start with
a zero as this may cause some computer programs to reject the
input. The code is based on the fact that the unit periods are
constant, and therefore, their sum may be set equal to zero. See
what effect this has on the normal equations for the straight line.
∑ y = na + b ∑ x
(c)

∑ xy = a ∑ x + b ∑ x2
If ∑ x = 0, the equations reduce to

∑ y = na
∑ x y = b∑ x2
 Quantitative Techniques for Management Applications

340
which allow the direct solution for a and b as follows

S
Notes

___________________ Σy
a=
n
___________________
Σxy
b= 2
___________________ Σx

E
___________________
This form simplifies the calculations substantially compared to the
___________________ previous formulae. The code, however, that allows to set å x = 0
___________________ must incorporate the integrity of a unit distance series. Thus, if the

UP
series is odd-numbered, the midpoint is set equal to zero and the
___________________
code completed by negative and positive unit distances of x=1,
___________________ where each x unit stands for one year or other time period. If the
___________________ series is even-numbered, let us say it ran from 2003 to 2012, the
two midpoints (2007/2008) are set equal to-1 and +1, respectively.
___________________
Since there is now a distance of x = 2 between + 1 (-1, 0, +1), the
code continues by negative and positive units distance of x = 2
where each x unit stands for one-half year or other time period.
The worksheet is in Table 24.3 and calculations are as follows:
E-
Table 24.3: Worksheet

YEAR Code for an YEAR Code for an Dividend XY X2

Even Series Odd Series payments
X X in ` Y
2003 -9
2004 -7 2004 -4 2.2 -8.8 16
2005 -5 2005 -3 2.4 -7.2 9
CC

2006 -3 2006 -2 3.0 -6.0 4

2007 -1 2007 -1 5.0 -5.0 1
2008 1 2008 0 6.8 0 0
2009 3 2009 1 8.1 8.1 1
2010 5 2010 2 9.0 18.0 4
2011 7 2011 3 9.5 28.5 9
2012 9 2012 4 9.9 39.6 16
Total 0 0 55.9 67.2 60

5.59
Then a= = 6.211
9
(c)

67. 2
b= = 1.12
60
and Yc = 6.211 + 1.12 x
origin 1995
x in 1 year units
UNIT 24: Forecasting

341
Note that in the case of time series analysis, the origin of the code

S
Notes
and the x units must be defined as part of the regression equation.
In our problem the investor would like to obtain a dividend ___________________
forecast for 2014. Since the origin is 2008 (x = 0) and x = 1 year ___________________
units, the code for 2014 is x = 6. Therefore the forecast is yc =

E
___________________
6.211+1.12 (6) = ` 12.9. If the time series had been even numbered,
___________________
let us say that dividend payments for 2003 had been included in
the forecasting study, the definition under the regression equation ___________________
would have read ___________________

UP
origin 2007/08 ___________________

x in 6 month units. ___________________

Thus, we know that for 2008, x=1; and since we must use x=2 units ___________________
for each year, the code value for 2014 would be x=13. Once the yc ___________________
value has been obtained, b is tested for significance and the 95%
confidence interval constructed as previously shown.
Time series analysis is a long-term forecasting tool. Hence, it
addresses itself to the trend component T in our time series
E-
equation TS = T + S. In the dividend forecast, b=1.120 was
calculated which means that in the environment that is reflected
in the set, Smart increased the dividend payments on an average
by ` 1.12 per year. Let us now turn out attention to the seasonal
variation component, that may be present in a time series. A
product’s seasonality is shown by the regularly recurring increases
CC

or decreases in sales or production that is caused by seasonal

influences. In the case of some products, their seasonality is quite
apparent. As an obvious example virtually all non-animal
agricultural commodities may be cited. Seasonality of other
products may be more difficult to detect. Take hogs in order to stay
on the farm. Are they seasonal? They are lusty breeders and could
not care less about seasonal influences. Yet, there is an induced
season by the corn harvest. If corn is plentiful and cheap, farmers
raise more hogs. This is known as the corn-hog cycle. Or take
automobiles, Indian manufacturers are used to introduce major
design or technological changes once every generation. This
(c)

“season” has now been shortened somewhat. How about

computers? There the season even has a special name. It is called a
generation and prior to increased competitive pressures within the
industry it used to be about seven years long. Our stock market
investor knows that stock trades on the Stock Exchanges are
 Quantitative Techniques for Management Applications

342 seasonal. The daily season is V-shaped starting the trading with a

S
Notes relatively high volume which tapers off toward the lunch hour to
___________________ pick up again in the afternoon. And so it goes with many other
products, not ordinarily thought of as being seasonal.
___________________

___________________ Let us quantify this seasonality and illustrate how it may be used

E
in a decision situation. There are, as is often the case, a number of
___________________
decision tools that may be applied. The reader may be familiar
___________________ with the term ratio-to-moving-average. It is a widely used method
___________________ for constructing a seasonal index and programs are available in

UP
larger computer libraries. Usually, the method assumes a 12 -
___________________
period season like the twelve months of the year. There is a more
___________________
efficient method which yields good statistical results. It is
___________________ especially helpful in manual calculations of the seasonal index and
___________________ when the number of seasonal periods is small like the four
quarters of a year, the six hours of a stock exchange trading day or
the five days of a work week. This method is known as simple
average and will be used for example purposes.
To stay with the investment environment of this unit section, let
E-
us calculate a seasonal index for shares traded on the Stock
Exchange from July 2 through July 7, 2012. This period includes
the July 4 weekend. Volume of shares (data) for each trading day
(season) is given in thousands of shares per hour. The Individual
steps of the analysis (operations) are discussed in detail for each
column of the worksheet below.
CC

Table 24.4: Worksheet

Column (3) (4) (5) (6) (7) (8) (9) (10)
(2)
Hour Total Trend Seasonal Seasonal 7/2 7/3 7/6 7/7 Avg.
Variation Variation Variation Index
(TS) (T) TS-T
10-11 0.965 0 0.965 110.6 12.00 12.25 15.44 16.72 14.10
11-12 0.245 0.159 0.086 103.7 10.40 11.75 15.04 16.32 13.38
12-13 -0.885 0.318 -1.117 94.2 10.55 10.06 12.95 15.44 12.25
13-14 -1.555 0.477 -2.032 87.1 9.55 9.46 12.05 15.24 11.58
14-15 0.395 0.636 -0.241 101.1 11.02 11.55 14.82 16.73 13.53
15-16 0.835 0.795 0.040 103.3 11.58 12.25 15.38 16.69 13.97
Average -0.383 600 10.85 11.22 14.28 16.19 13.135
(c)

As you inspect the data columns, you notice the V-shaped season
for each trading day. You also notice in the total daily volume that
there is an increase in shares traded. Hence, you can expect a
positive slope of the regression line. The hourly mean number of
shares is also indicated. This is the more important value because
UNIT 24: Forecasting

we are interested in quantifying a season by the hour for each 343

S
trading day. Now turn to the operations. In last column, the hourly Notes
trading activity for the four days has been summed. In this total ___________________
all time series factors are assumed to be incorporated. You will
___________________
recall that the positive or negative cyclical and irregular

E
component effect is assumed to cancel out over time. Hence ___________________

averaging the trading volume over a long term data set eliminates ___________________
both components, yielding TS=T+S. You may ask, are four days a
___________________
sufficiently long time span? The answer is NO. In a real study you
___________________
would probably use 15 to 25 yearly averages for each trading hour.

UP
In an on-the-job application of this tool, you will have to know the ___________________
specific time horizon in order to effectively eliminate cyclical and ___________________
irregular variations. But by and large, what is a long or short time
___________________
span depends upon the situation.
___________________
In order to isolate the trend component (T) so that it may be
subtracted from column (2) in the Table 24.4, yielding seasonal
variation, the slope (b) of the regression line must be calculated.
(Remember: b is T.) The necessary calculations are performed
E-
below using the mean hourly trading volume for each day. But
since we are interested in an index by the hour, the calculated
daily b value must be apportioned to each hour. This is
accomplished by a further division by six—the number of trading
hours. The result is entered in column (3). Note that the origin of a
time series is always zero. The origin of the time series is always
the first period of the season. In our case this is the 10-11 trading
CC

hour. Therefore the first entry in column (3) is always zero to be

followed by the equal (since this is a linear analysis) summed
increment of the apportioned b-value.

Table 24.5

Average Hourly
Day Code Trading Volume
Per Day
x x1 y xy x2
7/1 -3 10.85 -32.55 9
7/5 -1 11.22 -11.22 1
(c)

7/6 1 14.28 14.28 1

7/7 3 16.19 48.57 9
Total 52.54 19.08 20
 Quantitative Techniques for Management Applications

344
Σxy

S
Notes
Activity b= 2
Σx
What are the limitation of
___________________
linear analysis? 19.08
=
___________________ 20
___________________ = 0.954

E
___________________ and the apportioned b-value is
___________________
0.954
___________________ = 0.159
6

UP
___________________
It is not necessary to calculate the y-intercept (a) in this analysis
___________________ unless of course, you wish to combine it with a long-term forecast
___________________ of daily trading volume. Then, just to review the calculations, you
___________________
would find:

Σy
a=
n
52.54
=
E-
4
= 13.135
and

yc = 13.135 + 0.954 x

origin 7/5 and 7/6

CC

x in half trading day units.

In column (4) TS - T = S is performed. Column (4) is already a

measure of seasonal variation. But in order to standardize the
answer so that it may be compared with other stock exchange, for
example, it is customary to convert the values in column (4) to a
seasonal index. Every index has a base of 100 and the values above
or below the base indicate percentages of above or below “normal”
activity, hence the season. Since the base of column (5) is 100, the
mean of the column should be 100 and the total 600 since there are
6 trading hours. In order to convert the obtained values of column
(4) to index numbers, each of its entries is added to the total mean
(c)

and then is divided by the column mean added to total mean and
multiplied by 100 yielding the corresponding entry in column (5). It
is customary to show index numbers with one significant digit.
Column (6) shows the seasonal effect of this decision variable—
share trading on the Stock Exchange. Regardless of heavy or light
UNIT 24: Forecasting

daily volume, the first hour volume is the heaviest by far. It is 7.4% 345

S
above what may be considered average trading volume for any Notes
given day. Keep in mind that a very limited data set was used in ___________________
this analysis and while the season, reaching its low point between
___________________
1 and 2 p.m., is generally correctly depicted, individual index

E
members may be exaggerated. What managerial action programs ___________________

would result from analysis such as this? Would traders go out for ___________________
tea and samosas between 10-11? How about lunch between 1-2?
___________________
When would brokers call clients with hot or lukewarm tips?
___________________
Assuming that a decrease in volume means a decrease in prices in

UP
general during the trading day, when would a savvy trader buy? ___________________
When would he sell? Think of some other intervening variables ___________________
and you have yourself a nice little bull session in one of Dalal
___________________
Street’s watering holes. If, in addition, you make money for
yourself or firm, then, you have got it. ___________________

Check Your Progress

Fill in the blanks:
E-
1. Time series analysis is ________ term forecasting tool.
2. Product seasonablity is shown by regular occuring______
or _______ is sale

Non-linear Analysis
Any number of different curves may be fitted to a data set. The
CC

most widely used program in computer libraries, known as

CURFIT, offers a minimum of 5 curves plus the straight line. The
curves may differ from program to program. So, which ones are the
“best” ones? There is no answer. Every forecaster has to decide
individually about his pet forecasting tools. We will discuss and
apply three curves in this section. They appear to be promising
decision tools especially in problem situations that in some way
incorporate the life cycle concept and the range of such problems is
vast, indeed.
If you take a look again at Figure 24.5, you see that three curves
(c)

have been plotted. As we know from many empirical studies,

achievement is usually normally distributed. Growth, on the other
hand, seems to be exponentially distributed. The same holds true
for decline. As the life cycle moves from growth to maturity, a
parabolic trend may often be used as the forecasting tool. These
are two of the curves that will be considered. The third one is
 Quantitative Techniques for Management Applications

346 related to the exponential curve. As you look at the growth stage

S
Notes
Activity and mentally extrapolate the trend, your eyes will run off the page.
How___________________
does parabola fit is Now, we know—again from all sorts of empirical evidence—that
applied in managerial
trees don’t grow into the high heavens. Even the most spectacular
___________________
applications?
growth must come to an end. Therefore, when using the
___________________

E
exponential forecast, care must be taken that the eventual ceiling
___________________ or floor (in the case of a decline) is not overlooked. The modified
___________________ exponential trend has the ceiling or floor build in. It is the third
curve to be discussed.
___________________

UP
One final piece of advice before we start fitting curves. If you can
___________________
do it by straight line, do it. For obvious reasons, just look at Figure
___________________
24.5, any possible error—and there is always a built-in five percent
___________________ chance—is worse when a curve is fitted. By extending the planning
___________________ and forecasting horizon over a reasonable shorter period rather
than spectacular but dangerous longer period, the straight line can
serve as useful prediction tool.

The Parabola Fit

E-
The parabola is defined by
yc = a + bx + cx2

Where a, b and c are constants a and b have been dealt. c can be

treated as acceleration. The normal equations are (method of least
square).
Σy = na + bΣx +cΣx2
CC

Σxy = aΣx + bΣx2 + cΣx3

Σx2y = aΣx2 + bΣx3 + cΣx4

Setting Σx= 0 as previously explained, Σx3 will also be zero.

Σy = na + cΣx 2

Σx 2 y = aΣx 2 + cΣx 4
Σxy
b=
Σx 2
(c)

There are direct formulae for a and c as well, but because of the
possible compounding of arithmetic error in manual calculations, it
is safer to solve a and c algebraically in this case.
To illustrate the parabolic trend let us forecast earnings per share
in dollars for Storage Technology Corporation for the years 2014
UNIT 24: Forecasting

and 2015. Storage Technology manufactures computer data 347

S
storage equipment, printers, DVD-ROMS and telecommunication Notes
products. The company was founded in 1969 and after going ___________________
through a period of explosive growth seems to be moving into the
___________________
maturity stage. Data, code and calculations are shown below in the

E
usual worksheet format. ___________________

___________________
Code Earnings Per Share
Year xy x2 x2y x4 ___________________
(x) (y)
___________________

UP
2007 -3 0.39 -1.17 9 3.51 81
___________________
2008 -2 0.54 -1.08 4 2.16 16
___________________
2009 -1 1.13 -1.13 1 1.13 1
___________________
2010 0 1.58 0 0 0 1
___________________
2011 1 1.72 1.72 1 1.72 1

2012 2 2.50 5.00 4 10.00 16

2013 3 1.84 5.52 9 16.56 81

Total 0 9.70 8.86 28 35.08 196

E-
8.86
Then b =
28
= 0.3164
and solving simultaneously
9.70 = 7a + 28c × 4
CC

⇒ 38.8 = 28a + 112c

35.08= 28a + 196c
3.72 = 84c
c = 0.0443
a = 1.2085
Therefore,

yc = 1.2085 + 0.3164 x- 0.0443x2

origin 1996
(c)

x in 1 year units
and specifically,

y2014 = 1.2085 + 0.3164(5) - 0.0443(5)2

= ` 1.68,
 Quantitative Techniques for Management Applications

348
y2015 = 1.2085 + 0.3164(6) -0.0443(6)2

S
Notes
Activity
What is the scope of
___________________ = ` 1.51
exponential fit?
___________________ Remember that the data set is small. Quarterly earnings per share
___________________
figures for the period may have been better because of the larger

E
sample size. The significance test and construction of the
___________________
confidence interval is performed as previously shown.
___________________ Furthermore, as soon as new earnings per share figures become
___________________ available, the regression line should be re-calculated, because

UP
there is always the chance that there may be a change in the
___________________
environment.
___________________

___________________ The Exponential Fit

___________________ This illustrative forecasting study is performed for the Acme
Company Ltd. that manufactures toy rubber ducks to be used in
bathtubs. Over the past few quarters, the company has become a
major defence contractor. The Navy is buying an ever increasing
number of the ducks as part of its rearmament program. Shipment
E-
figures are kept secret to confuse the enemy—and the media.
Therefore, the data in the accompanying table are hypothetical.
We may fit an exponential trend which takes the form
yc = abx

As previously mentioned, exponential trends are difficult to plot,

because you run very quickly off the top of the page. However,
CC

when using semi-log paper (the y-axis is in logarithmic scale), the

trend appears as a straight line. This phenomenon may be used to
good advantage when calculating a and b.
Thus using the logarithmic form of the exponential trend
log yc = log a + x log b

The straight line equations may be used, or,

Σ log y
log a =
n
and
(c)

Σ log y
log b = 2
Σx
UNIT 24: Forecasting

349
when Σx = 0. The data set and calculations appear in the

S
Notes
worksheet below. Logarithms are obtained from a pocket calculator
or any standard table. ___________________

___________________
Quarter
Shipments
Since

E
Code in Thousands ___________________
Initial log y x log y x2
(x) of Units
Navy ___________________
(y)
Contract
___________________
1 -5 2 0.30103 -1.50515 25
2 -3 4 0.60206 -1.80618 9 ___________________

UP
3 -1 9 0.954243 -0.954243 1 ___________________
4 1 20 1.30103 1.30103 1
___________________
5 3 55 1.740363 5.221089 9
6 5 110 2.041393 10.206965 25 ___________________
Total 0 7.241149 12.463511 70 ___________________

7.241149
Then log a =
6
= 1.2069
E-
12.463511
and log b =
70
= 0.1781
when expressed in logarithmic form. The regression equation is
log yc = 1.2069 + 0.1781 x
CC

origin third and fourth quarters

x in half quarter units.
The equation may be used in this form for forecasting purposes.
Suppose that Acme would like to have a forecast for the next two
quarters.
The forecasts are
log y7 = 1.2069 + 0.1781(7)

= 2.4536 and finding the anti-log

y7 = 284 thousand more rubber ducks
(c)

and
log y8= 1.2069 + 0.1781 (9)

= 2.8098 and finding the anti-log

y8 = 645 thousand still more rubber ducks
 Quantitative Techniques for Management Applications

350
If you transform the logarithmic form of the regression equation,

S
Notes
there is something interesting to be seen if you remember the
___________________
compound interest formula. It works like this-
___________________
log yc= 1.2069 + 0.1781x and finding the anti-log
___________________

E
yc= (16.1) (1.51)x
___________________

___________________ which may be rewritten

___________________ yc= (16.1) (1+0.51)x,

UP
___________________ And you recognize that it takes the form of the compound interest
___________________ formula where the rate is 0.51 or 51%. This is Acme’s average
quarterly increase in its defence business.
___________________

___________________ The Modified Exponential Fit

In any case other than military procurement, except in those
countries that have bled themselves dry because of it and now have
neither the money for military extravaganza nor civilian
necessities/amenities, trees don’t grow into the high heavens.
E-
Given this profound observation, there must be a decision tool that
places a cap or ceiling on overly exorbitant growth forecasts. But
they may want to consider the other option: exponential economic
declines do not always result in a merciful state of down and out
but gradually approach a floor which may be called subsistence,
making do, squeezing by or other nice and flowery allegories. At
CC

any rate the asymptote of the modified exponential curve is,

yc = k + abx,

Where k is the asymptote, provides us with such a tool. There are

four cases.
(c)

Figure 24.7
UNIT 24: Forecasting

A least squares fit is not efficient in this case. Rather a solution 351

S
method is discussed that is based on the theorem that the ratio of Notes

successive first differences between points on the exponential ___________________

curve is constant and equal to the slope b.
___________________
The decision tool is known as the method of semi-averages. It is

E
___________________
based on the calculation of three sums of successive points of the
___________________
time series. Therein lies the limitation of this technique, because
the number of data points must be divisible by three. Thus, a ___________________

minimum of six points is necessary and if the time series consists ___________________

UP
of n=20 data points, the two earliest one (to preserve the most
___________________
relevant environment) must be eliminated. The formulae for a, b
___________________
and k are generated as follows from six general y-values starting
with the origin of the series. ___________________

y0 = k + a ⎫
___________________
⎬ Σ1 = y0 + y1
y1 = k + ab ⎭

y2 = k + ab2 ⎫
⎬ Σ2 = y2 + y3
y3 = k + ab3 ⎭
E-
y4 = k + ab4 ⎫
⎬ Σ3 = y4 + y5
y5 = k + ab5 ⎭

then

Σ1 = 2k + a(b+1)
CC

Σ2 = 2k + ab2 (b+1)

Σ3 = 2k + ab4 (b+1)
and

2 Σ3 - Σ2
b =
Σ 2 - Σ1
Σ 2 - Σ1
2k = Σ 1 - 2
b -1
b -1
a = (Σ 2 - Σ1 )
(c)

2 2
(b - 1)( b - 1)

or, in the general case involving a time series of n data points and
n is divisible by three.
 Quantitative Techniques for Management Applications

352
Σ3 - Σ2

S
Notes n
b =
___________________ Σ 2 - Σ1
Σ 2 - Σ1
___________________ nk = Σ 1 - n
b -1
___________________ b -1

E
a = (Σ 2 - Σ 1 ) n n
___________________ (b - 1)( b - 1)
___________________
Suppose a set consists of the following data points.
___________________
Year Code Sales Units

UP
___________________ 2007 0 100
⎫
___________________ ⎬ Σ1 = 260
2008 1 160 ⎭
___________________ 2009 2 200
⎫
___________________ ⎬ Σ2 = 430
2010 3 230 ⎭
2011 4 245
⎫
2012 5 250 ⎬ Σ3 = 495
⎭

Then
E-
2 495 - 430
b =
430 - 260
= 0.38
b = 0.62
CC

430 - 260
2k = 260 -
0.38 - 1
= 260 - (-274.19)
= 534.19
k = 276.10
0.62 - 1
a = (430 - 260)
(6.38 - 1)(0.38 - 1)
= (170) (- 0.99)
= -168.3
(c)

and
yc= 267.10 + (-168.3) (0.62)x

which makes it a Case 4 curve with a sales ceiling of 267.10. The

forecast is made in the usual manner. For 2013 it is
UNIT 24: Forecasting

353
y2013 = 267.10 +(-168.3) (0.62)6

S
Activity
Notes
= 267.10 +(-9.56) What is the limitation of
___________________
modified exponential fit?
= 257.54 units. ___________________

E
___________________
Check Your Progress
___________________
Fill in the blanks
___________________
1. The decision tool is known as method of _________ .
___________________
2. Shipment figures are kept secret to confuse the

UP
__________ and the ___________ . ___________________

___________________
Multiple Regression Analysis ___________________

At this point we have a fairly good understanding about how to ___________________

approach, a forecasting problem with one independent variable.
We know how to fit a straight line or three widely applicable
curves to a data set. But even as we generated the functional
relationship of our first problem—Sales= f (Advertising) — the
E-
thought must have occurred to us if advertising were the only
predictor variable in this case. And as you can sense, the decision
situation will be considerably broadened thereby. It is not only a
matter of incorporating more variables into the decision space, or,
mathematically speaking the move from y=f(x) to y = f (x1, x2,...,xn)
and in the form of the regression equation from y=a+bx to yc = a +
b1 x1 + b2 x2 + ...... + bn xn, but rather by testing additional variables
CC

effect and then by substituting another one, we are able to

simulate various environments. In this sense multiple regression
analysis may not only be a more sophisticated forecasting tool, but
it can nicely serve to sharpen the decision maker’s understanding
of the forecasting environment and thus, serve as a managerial
training device.

To get us started in this new light of somewhat more

sophistication, a few background review comments may be in
order. The relationships between independent and dependent
variables in the multiple regression problems are assumed to be
linear. The methodology for a non-linear multiple regressions have
(c)

not really advanced beyond the research stage. This does not mean
that inherently non-linear variables cannot be accommodated
within the analysis system, but they must be transformed. The
growth situation of the exponential trend problem comes to mind.
Remember that we then introduced linearity back into the picture
by using the logarithm of the variable’s data—y in that case. We
 Quantitative Techniques for Management Applications

354
transformed the data to make them appear linear. Other

S
Notes techniques than logarithmic transformation are available.
___________________ The fitting technique for multiple regressions is again the least
___________________ squares method which is based on the Gauss-Markov Theorem. It
holds that when the population variables are fixed and the Y-
___________________

E
variable random, the variances of the sub- population ys for
___________________ corresponding xs are equal, and the probability of the system being
___________________ in a certain state (environment) initially is the same probability
throughout the system, the best estimation of the population
___________________
regression y = a + b x by the sample regression yc = a+bx is the

UP
___________________ method of least squares. In Figure 24.1 we noted the deviation y-yc,
___________________ which is known as stochastic disturbance or simply error (e). You
remember that the method of least squares minimizes the sum of
___________________
the squared error. Further, it is assumed that the effort is
___________________ normally distributed, hence y is normally distributed and
therefore, we can test a and b by t-test after estimating
s a and s b. We did only the latter, but soon you will see computer
printouts that include the former as well.
Finally a few comments may be made about operational aspects.
E-
The normal equations are as follows for the multiple regression
problem yc= a + b1 x1 + b2 x2 + b3 x3
ΣY = na + b1 Σx1 + b2 Σx2 + b3 Σx3

Σx1y = aΣx1 + b1 + b2 Σx1 x2 + b3 Σx1 x3

Σx2y = aΣx2 + b1 Σx2 x1 + b2 + b3 Σx2 x3

CC

Σx3y = aΣx3 + b1 Σx3 x1 + b2 Σx3 x2 + b3 Σx3

This is the general system of equations and we will consider an

actual problem shortly. By comparing the normal equations for the
straight line with the ones above, you can readily see how they
were developed and you can easily generate the equations for 10 or
15 variable problems as frequently encountered in econometrics.
What is not so easy, indeed it may be impossible, is to solve this
many equations simultaneously. Here matrix algebra helps. But
even then manual calculations are forbidding. Therefore, multiple
regression analysis must be performed by computer and we will
not even bother with a simplified manual example as before.
(c)

However, so that we may have an inkling about the computational

procedures, let us take a look at the multiple correlation formula
for a problem with two independent variables
r122 + r132 + 2 r12 r13 r23
R123 =
1 - r232
UNIT 24: Forecasting

355
Where R and r are the multiple and simple correlation coefficient,

S
Notes
respectively, and y=1, x1 = 2 and x2 = 3. As you can see the
procedures rest on performing iterative simple correlation ___________________

analysis. This, of course, is what computers love to do. Managers ___________________

don’t.

E
___________________
Having solved this operational problem, let us look at the second ___________________
and trickier, one. It has to do with a relaxation of the underlying
___________________
assumptions of regression analysis. Such relaxation may be
necessary because of aspects that exist in the environment. First, ___________________

UP
it is assumed that the error (e) terms a, b, g, d are independent, or ___________________
re = 0. In trend analysis, however, a long term increase or decrease
___________________
virtually assures dependency. This is known as serial or auto
___________________
correlation. Tests and significance tables have been prepared. The
Durbin-Watson statistic, soon to be encountered in our computer ___________________

analysis, allows the identification of serial correlation. Secondly, it

is assumed that the independent variables x1, x2, ..., xn are
mutually independent. If this assumption is relaxed, sb tends to
E-
become large resulting in an insignificant b. Small potatoes, as you
recall. This problem is known as multi-collinearity, and techniques
are available to test for its existence. Remember, however, that the
logic of a well-designed conceptual model should identify the
problem already at the stage of forecasting. If it exists, there are
only two ways out: get additional data, or more typical, get other
variables.
CC

There are three more, though minor, problems that may arise
because of the underlying assumptions. The first one was
mentioned already and is only repeated here. The assumption is
made of linearity or constant variance s2. It has the name of
homoscedasticity. We already know that the behaviour of some
variables tends to result in non-linearity and that some “linear”
variables — as shown in Figure 24.5— may display non-linearity
at certain magnitudes. It was suggested to use transformations in
those cases. This is sound advice as long as the forecaster knows
which way s2 varies. If unknown, the transformation cannot be
(c)

performed. The second problem involves lagged variables such as

in
yc = a + b1 x b + b2 xt–1

Where xt is the x value of time period t and xt-1 the x value of the
previous time period. We will encounter lagged variables again in
 Quantitative Techniques for Management Applications

356 the later discussion about smoothing. Indeed, since lagged

S
Notes variables may contain both multicollinear and serial correlation,
___________________ perhaps the initial forecasting method should rely on smoothing.
There are at present no meaningful methods to cope with this
___________________
problem in the regression framework.
___________________

E
Finally, there is the general assumption governing the whole body
___________________
of quantitative analysis, namely, the variables must be
___________________ quantifiable. We said before rather non-chalantly that this would
___________________ not pose a problem. May be not, but, as you may have noticed, our
discussion is moving away from the conceptually simple decision

UP
___________________
problems into the more complex ones. In the case of the first type,
___________________ once the problem has been defined and a solution method
___________________ algorithm explained, the situation is clear. The second type calls
for extensive model building abilities and training. Here the
___________________
problem and algorithm are clear but not necessarily how to
quantify complex variables and to show their interrelationships.
There are three qualitatively different types of variables in any
decision situation. They may be called concrete, abstract and
E-
nonsense. Concrete variables can be measured easily. Take heat,
we obtain a physical measurement in centigrade. Take advertising
expenditure, we obtain a dollar or rupee measurement. Now take
beauty which, as anyone knows, is in the eye of the beholder. This
is an abstract variable. But it is easily quantifiable by rankings as
any beauty contest demonstrates. The same holds true for a
nonsense variable concept as found in many product names. But
CC

now suppose that one of the variables in a sales forecasting study

is gender. Let us say that it is known that women buy more of the
product than men. In this case, data collection procedures assign a
1 when women respond and a 0 for the response of men. Such
variables are known as dummy variables and, as of now, this term
has been removed from your catalog of insults. Make it part of the
model building knowledge. Here is how it works.
Take
yc = a + b1 x1 + b2 x2

Where yc is sales revenue, b1 advertising expenditures and b2

(c)

gender. In this equation yc and x1 are quantitative and x2

qualitative in nature. Now setting
x2 = 1, when a women responded

x2 = 0, for responses from all others,

UNIT 24: Forecasting

357
we obtain the following equations

S
Notes
y c = a + b1 x 1 for “all other response”
___________________
yc = (a + b2) + b1 x1 for “woman response”. ___________________
You notice that the slope of the regression lines is b1, but that a the

E
___________________
y - intercept differs. In the case of “Woman response”, the line ___________________
shifts upward, parallel to the all other responses line. Similarly
___________________
regression equations may be solved when all independent or the
dependent variables are qualitative in nature. The disadvantage of ___________________

UP
the dummy variable, however, is the fact that it can take only the 0 ___________________
or 1 value. There are methods that allow qualitative rankings with
___________________
more than 2 response choices like agree, don’t know, disagree.
___________________
Such a variable is known as polychotomous and just from the term
you know that it is beyond the scope of this book. But if you ever ___________________
meet such a variable in some dark decision situation, brush up on
multivariate analysis. That allows you to handle this stranger.
Now let us look at an old friend in new clothes and demonstrate
E-
how multiple problems are actually solved. From now on it is
strictly computer work and every computer library has one or more
MULREG programs, as they are often called. (To use the most
used problems we will use ‘Regression Analysis’ in ‘Tools’ section of
‘Microsoft Excel 97’). The problem is the same that we solved for
our controller via simple linear regression except that we change it
from Sales = f (Advertising) to Sales = f (Advertising, Percent
CC

spendable income allocated to product type by our customers as

shown by a market research study, Time). Perhaps we thought
that an analysis like this would put us on the promotional fast-
track of the company. We shall see. First the data set.

Sales Advertising Per cent Time

` lacs ` lacs Allocated Year Code

y x2 x3 x4

30 2.8 13.8 1995 1

40 4.2 12.2 1996 2
55 5.0 15.0 1997 3
(c)

50 5.5 16.0 1998 4

70 7.5 18.0 1999 5

Let us mention the caveat again: this data set lands us in the dog
house rather than on the fast-track because of its paucity. But we
 Quantitative Techniques for Management Applications

358 are interested only in an example of the procedures. Note that the

S
Notes dependent and three independent variables are called y and x
___________________ respectively and x variable are numbered consecutively. This is for
data input purposes. Thus the multiple regression equation is
___________________

___________________ y = a + b2 x2 + b3 x3 + b4 x4.

E
___________________ Each computer system calls for slightly different instructions and
data input for the Mulreg program. By looking carefully at the
___________________
'Microsoft Excel 97' procedures below, you can readily infer how
___________________
your system works.

UP
___________________
Before we look at them let us explain the reasons for this analysis.
___________________ Remember that multiple regression analysis is as much as
___________________ forecasting tool as it is a managerial training device. Obviously we
emphasis the training aspect here. First we want to perform the
___________________
simple regression analysis by using 'Regression Analysis' in
'Microsoft Excel 97' and you will see how quickly we can verify the
results to the manual calculations that we suffered through a
number of pages ago. So the first regression that we perform is
E-
simple and reads
y = f (x2)

Next we want to do the whole Schemer. You get

y= f (x2,x3,x4)

and you see how nice and easy life is for the quantitative decision
CC

maker.
The third regression is a time series analysis, or,
y = f (x2, x3)

and
y = f (x3, x4)

As you realize, by now, in this little training exercise we have

incorporated all the regression options that have been discussed so
far except the dummy variable. Its inclusion, let us say that the
product was seasonal and we had differentiated between “high
(c)

season” and “all others”, would pose no problem, except that then
we would have had to change the sample size and could not have
used the original problem. It works like this. In this multiple
regression analysis

df = n - m ≥ 1
UNIT 24: Forecasting

359
or, to rewrite this assumption in terms of the bs

S
Notes
n>k+1
___________________
where k is the number of bs or variables in the equation.
___________________
Otherwise a degenerate model results and that’s why the dummy

E
___________________
variable was not included. You may remember this point when
you run your own multiple regression problem with 15 variables. ___________________

In order to perform these five regressions, the instructions about ___________________

the 'Regression' dialog box are shown in Exhibit 24.1. ___________________

UP
Exhibit 24.1: About the Regression Dialog Box ___________________

Input Y range ___________________

Enter the reference for the range of dependent data. The range ___________________
must consist of a single column of data. Here the sales data is the ___________________
Y range.

Input X Range

Enter the reference for the range of independent data. Microsoft

Excel orders independent variables from this range in ascending
E-
order from left to right. The maximum number of independent
variables is 16. In this question the inputs have to be specified for
each of the regression calculation separately.

Labels

Select if the first row or column of your input range or ranges

contains labels. Clear if your input has no labels; Microsoft Excel
CC

generates appropriate data labels for the output table.

Confidence Level

Select to include an additional level in the summary output table.

In the box, enter the confidence level you want applied in addition
to the default 95 per cent level.

Constant is Zero

Select to force the regression line to pass through the origin.

Output Range

Enter the reference for the upper-left cell of the output table.
(c)

Allow at least seven columns for the summary output table, which
includes an ANOVA table, coefficients, standard error of y
estimate, r2 values, number of observations and standard error of
coefficients.
Contd….
 Quantitative Techniques for Management Applications

360 New Worksheet Ply

S
Notes Click to insert a new worksheet in the current workbook and past
___________________ the result starting at cell A1 os the new worksheet. To name of
new worksheet, type a name in the box.
___________________
New Workbook
___________________ Click to create a new workbook and past the results on a new

E
___________________ worksheet in the new workbook.
Residuals
___________________
Select to include standardized residuals in the residuals output
___________________ table.

UP
___________________ Standardized Residuals
___________________ Select to include standardized residuals in the residuals output
table.
___________________
Residual lots
___________________ Select to generate a chart for each independent variables versus
the residual.
Line Fit plots
Select to generate a chart for predicted values versus the observed
values.
E-
Normal Probability Plots
Select to generate a chart that plots normal probability.

After we have discussed each regression, it is suggested that you

run this very problem with your computer program unless you are
already familiar with it.
Regression 1 (y = a + b2 x2).
CC

The printout for this simple regression problem is shown in

Exhibit 24.2. It was previously solved manually.

Exhibit 24.2: Summary Output of Regression of

Sales (y1) and Advertising (x2)
(c)
UNIT 24: Forecasting

Let us proceed step-by-step through the printout. The correlation 361

S
coefficient value shows a minor rounding error between the Notes
manual and computer solutions (r= 0.9716 and r = 0.9715784, ___________________
respectively). The Anova matrix is of no concern to us here. This
___________________
matrix is followed by the values of the regression equation where

E
coefficient indicates the y - intercept (a) value and the other X ___________________

Variable 1 value the b value that corresponds to the numbers of ___________________

the independent variables. Thus, yc = 6.429 + 8.514x, as previously
___________________
obtained. Note the Standard. Error and T-STAT. The latter is
___________________
rounded t = 7.11 for b and as previously calculated.

UP
___________________
Let us now discuss the main regression statistics, the
R-squared and R-values. These are the adjusted values for the ___________________

multiple coefficients of determination and correlation, respectively. ___________________

These coefficients carry a positive bias in unadjusted form, that is, ___________________
an unadjusted R2 obtained by the formula results in an
overestimate of the true R2. In order to find the true R21 the
following adjustment formula is used
E-
2 2 n-1
RADJ = 1 - (1 - R )( )
n-m
Since in this case of a simple correlation problem R2 =
r2 = (0.9715782)2. Then

2 5-1
RADJ = 1 - (1 - 0.9439641)( )
CC

5-2
= 0. 9252855

while the printout shows 0.925286105. It should be noted that the

relatively large difference comes between the adjusted and
unadjusted R values, it is due to the small sample size. With a
2

meaningful sample size, the difference between R2 and R2ADJ is

usually negligible.
Regression 2 (y = a + b2 x2 + b3 x3 + b4 x4)

Now we are entering the realm of the multiple.

(c)

Notice the regression equation

y = yc = -8.3958 - 3.9583 x4 +0.9375x3 +11.0416 x2

and the T-STAT that is insignificant in every case. R2 ADJ

has
dropped from 92.5% of the variation explained (Regression 1) to
80.4%. Obviously, the additional variables did not give us a better
 Quantitative Techniques for Management Applications

362 forecasting tool. With an analysis like this, we won’t make the

S
Notes fast-track.
___________________
Regression 3 (y =a + b4 x4).
___________________
The b is significant again and R2 ADJ
has improved. Overall it does
___________________

E
not look as good as Regression 1 as a forecasting tool. Unless, after
___________________ elimination of the serial correlation, its R2 ADJ represents an
___________________ improvement over this value in Regression 1, we might as well
___________________
stay with Regression 1.

UP
___________________ Regression 4 (y = a + b3 x3 + b2 x2)

___________________ Well, there is a flicker of hope in Table 24.6.

___________________ R2 ADJ
has improved to 89% from 84% in Regression 3. But look at
___________________ the T-STAT of b3 which indicates an insignificant value and b2 just
makes it by one-tailed test with df = 2, or MEV = 2.92 according to
the t table. Conclusion: no improvement over Regression 1.
It shows you very nicely the interface between the conceptual and
E-
mathematical models as you can test any linear variable or
combination of variables keeping in mind the time and cost
constraints that apply to data collection. It also teaches you to
differentiate between statistical and operational significance which
helps you to sharpen the conceptual model building skills—the
most important asset of the forecaster.
CC

Table 24.6: Summary Output of Regression of Sales (y) with Advertising

(x2) and Percent Allocated (x3)
(c)
UNIT 24: Forecasting

363
Check Your Progress

S
Notes
Fill in the blanks:
___________________
1. There are _________ qualitatively different types of
___________________
variables in any decision situation.

E
___________________
2. Variables must be _____________
___________________

The Smoothing Algorithm ___________________

___________________

UP
If you have understood the basic aspects of forecasting up to this
___________________
point and are reasonably certain that you can use the regression
algorithm, you may not want to take this final section of the unit ___________________
too seriously. Read it, nevertheless, to know how the simple ___________________
minded folk of forecasters spend its days. Indeed, many authors
___________________
feel compelled to point out when introducing the subject of
smoothing that it is used when (a) “real” forecasting would be
prohibitively expensive (b) “real” forecasting talent has not found
its way into an organization’s staff (c) “real” forecasting talent is
not necessary because in the case of the operations or firms
E-
involved, it wouldn’t amount to much anyway and so forth. You
remember the comment about how simple minded decision making
folk knows how to phrase its utterances to make them sound
“scientific”? Well, this is it. Final introductory advice: Beware of
smooth smothers.

What do such types do anyhow? For some psychological reason

CC

they do not like the charming 1 and revealing little wiggles and
waggles (scientific: peaks and troughs) of time series and smooth
them out. They throw away the good stuff and eat chaff. How do
they do it? They average them out. But that is all. No further
sophistication is visible. Remember, they are simple minded folk.
Here is how it works. Suppose that you have the stock closing
prices of Joy Manufacturing Co. for the month of April, 1999.
During this month there were 21 trading days with their inevitable
wiggles and, unfortunately, waggles. By calculating a moving
average, they may disappear. The moving averages are known as
the smoothed values and are obtained by
(c)

x t-m/2 + x t-m/2+1 + x t +...+x tx-m/2+1 + x tx-m/2

st =
n

Where St is the smoothed value for time period t, m a movement

specification that is to be determined, x the data, and n the
number of adjacent data points that are to be averaged.
 Quantitative Techniques for Management Applications

364
Furthermore m is defined by m = (n-1)/2. Nobody will argue that

S
Notes this does not look very scientific, indeed.
___________________
Let us do it for a five-day moving average. The m = (5-1)/2=4
___________________ which means that four data points are lost. That is, two daily
___________________ stock prices at the beginning of the series and two at the end.

E
Hence
___________________
x1 + x 2 + x 3 + x 4 + x 5 26.25 + 27.25 + 28 + 28.125 + 28.25
___________________ s3 = = = 27.575
5 5
___________________ x 2 + x3 + x4 + x 5 + x6 27.25 + 28 + 28.125 + 28.25 + 28.375

UP
s4 = = = 28
___________________ 5 5
x 3 + x 4 + x 5 + x6 + x7 28 + 28.125 + 28.25 + 28.375 + 28.25
___________________ s5 = = = 28.2
5 5
___________________
x t-m/2 + x t-m/2+1 + x t +...+x tx-m/2+1 + x tx-m/2
___________________ st =
n

Where St is the smoothed value for time period t, m a movement

specification that is to be determined, x the data, and n the
number of adjacent data points that are to be averaged.
E-
Furthermore m is defined by m = (n-1)/2. Nobody will argue that
this does not look very scientific, indeed.

Let us do it for a five-day moving average. The m = (5-1)/2=4

which means that four data points are lost. That is, two daily
stock prices at the beginning of the series and two at the end.
Hence
CC

x1 + x 2 + x 3 + x 4 + x 5 26.25 + 27.25 + 28 + 28.125 + 28.25

s3 = = = 27.575
5 5
x 2 + x3 + x4 + x 5 + x6 27.25 + 28 + 28.125 + 28.25 + 28.375
s4 = = = 28
5 5
x 3 + x 4 + x 5 + x6 + x7 28 + 28.125 + 28.25 + 28.375 + 28.25
s5 = = = 28.2
5 5

and so on. The lagging is necessary because of St-1 which obviously

does not exist at X t-1. As you notice by doing a few more St values
manually, exponential smoothing uses the weighted average of
past time series values in order to compute the smoothed values
(c)

and it assigns greater weight to the most recent time series values.
The method assumes no significant long-term trend or seasonal
variation in the data. So what kind of a forecasting tool is this? As
pointed out before, if smoothing is to be used at all, it should be
used over the very short term only. Perhaps a trader, as opposed to
investor, may base stock purchases or sales on exponential
UNIT 24: Forecasting

smoothing. But check the effectiveness of this decision tool yourself 365

S
in the cases of the given two common stocks. Disregarding intra- Notes
day trading highs and lows, how often would our trader have been ___________________
successful in buying or selling the stocks of the companies if he had
___________________
bid the estimated price for each of the trading days? Then calculate
his profits or losses. The proof is in the pudding.

E
___________________

___________________
Check Your Progress
___________________
Fill in the blanks:
___________________
1. __________is management’s most important task, but

UP
not many managers in the public or private sectors are ___________________
good forecasters. ___________________

2. It is final introductory advice: __________ of smooth ___________________

smothers.
___________________

Case 24.1: There are 2117 Smart stores at petrol stations in

the India (the chain is building up). At present Smart has
reached an “upgrading” phase like so many discounters
before.
E-
Given the data below, perform the indicated analysis.
YEAR 1999 1998 1997 1996 1995 1994 1993 1992 1991 1990
EARNINGS
PER SHARE 19.0 17.5 20.7 28.4 27.4 23.9 21.1 16.1 8.5 11.1
DIVIDENDS 9.9 9.5 9.0 8.1 6.8 5.0 3.0 2.4 2.2 1.9
PER SHARE
CC

PRE-TAX 2.1 2.0 3.1 4.9 5.4 5.7 5.8 5.8 3.3 5.3
MARGIN

1. To what extent does the Board of directors regard dividend

payments as a function of earnings? Test whether there is a
significant relationship between the variables. Use a
parametric analysis.
2. Find the linear forecasting equation that would allow you to
predict dividend payments based on earnings and test the
significance of the slope.
3. Is there a significant difference in pre-tax margin when
comparing the periods 1995-1999 and 1990-1994. Perform a
(c)

non-parametric analysis. Explain the managerial implications

of your findings.
 Quantitative Techniques for Management Applications

366
Summary

S
Notes
There are two methodologies to anticipate future. They are called
___________________ qualitative and quantitative. The quantitative decision maker
___________________ always considers himself or herself accountable for a forecast-
within reason.
___________________

E
The qualitative school has generated many philosophical, religious
___________________
or political conceptual models according to which the ideology and
___________________ dogma is structured and forecasts prepared. In the guesstimate
___________________ conceptual model the forecast is based on expert opinion.

UP
___________________
The Delphi method consists of a panel of experts and a series of
rounds during which forecasts are made via questionnaire. The
___________________
second conceptual model stresses the fundamentals that impinge
___________________ upon the environment at any given time. In this case the forecaster
___________________
tries to ascertain the functional relationships among variables
defining the environment.
The mathematical models play a very important role in
forecasting. The quantitative analysis that underlies a forecast is
based on the type of conceptual model that has been chosen. Each
E-
of the two forecasting methods – regression and smoothing – has
their distinct algorithms. The forecast is nothing but an
extrapolation of the past and present. As previously pointed out, if
no fundamental changes take place, with respect to the magnitude
of the functional relationships among the variables and no new
variables enter the decision space, this logic has been found to
work well when applied.
CC

When the bivariate normal distribution cannot be assumed or the

observed data represent rankings, a non-parametric correlation
analysis must be performed. Correlation analysis examines the
functional relationship between dependent and independent
variables.

Lesson End Activity

Glass company is headquartered in Indore, M.P. It manufactures
glass and plastic storage jars for petrochemical industry.
Given the data below, perform the indicated analysis.
(c)

YEAR 2000 1999 1998 1997 1996 1995 1994 1993 1992 1991
EARNINGS
PER SHARE 1.25 2.89 2.81 3.09 3.52 2.97 2.71 2.13 1.59 1.77

DIVIDENDS 1.36 1.36 1.28 1.20 1.07 1.00 0.90 0.80 0.72 0.72
PER SHARE

PRE-TAX 1.1 4.6 5.2 5.5 8.9 7.2 10.1 7.9 7.3 9.7
MARGIN%
UNIT 24: Forecasting

367
(1) To what extent does the Board of directors regard dividend

S
Notes
payments as a function of earnings? Test whether there is a
significant relationship between the variables. Use a ___________________
paramedic analysis. ___________________

(2) Find the linear forecasting equation that would allow you to

E
___________________
predict dividend payments based on earnings and test the
___________________
significance of the slope.
___________________
(3) Is there a significant difference in pre-tax margin when
___________________
comparing the periods 1996-2000 and 1991-1995? Perform a

UP
parametric analysis. ___________________

___________________
Keywords
___________________
Intervening Variables: Those variables that may affect the
___________________
dependent variable's magnitude but which are not considered in
the decision space of the forecast either because of oversight or
their effect is deemed negligible, are known as intervening
variables.
E-
Correlation Coefficient: The degree of correlation between
normally distributed dependent and independent variables is
signified by the correlation coefficient.

Questions for Discussion

1. Define forecasting. What are the key methodologies of
CC

forecasting?

2. State the difference between the conceptual and mathematical

model of forecasting.

3. Discuss the concept of algorithms and their applications.

4. How will you conduct a simple linear regression analysis?

5. Write a note on time series analysis.

6. Illustrate the concept of multiple regression analysis.

7. Employment figures in thousands for Neo-Classical City and

(c)

suburbs are given below. Perform the required analysis.

(a) Using linear forecasts, predict the year in which

employment will be the same for the two locations.

(b) Construct the NCC confidence interval for that year.

 Quantitative Techniques for Management Applications

368
(c) Correlate the employment figures for the two areas using

S
Notes
both parametric and non-parametric methods and test the
___________________
significance of the correlation coefficients.
___________________
(d) Fit a modified exponential trend to SUB data and discuss
___________________
the results in terms of your findings in (1) above.

E
___________________
(d) Are NCC employment figures uniformly distributed over
___________________ the period 1994 through 2000?
___________________ YEAR 1994 1995 1996 1997 1998 1999 2000

UP
___________________ NYC 64.1 60.2 59.2 59.0 57.6 54.4 50.9

___________________ SUB 20.7 21.4 22.1 23.8 24.5 26.3 26.5

___________________ 8. Shown below are data sets that have been compiled by the
___________________ Reserve Bank of India and The Department of Commerce. All
amounts are in billions of rupees. Perform the following
analysis:

(a) Fit a modified exponential trend to Other Checkable

Deposits such as NOW accounts and predict the 2002
E-
value. Compare this value to the actual one.

(b) Is there a significant difference in the Percent Cash

Purchases when comparing the first half of the series
against the second half?

(c) Predict Personal Consumption on the basis of Consumer

Credit in the amount of ` 500 billion and test the
CC

significance of the slope b.

(d) Predict Demand Deposits for 2002 by linear trend.

Year Demand Other Consumer Personal Percent
Deposits Checkable Credit Consumption Cash
(` bn) Extensions Expenditures Purchases (` bn)
(` bn) (` bn) (` bn)

1990 53.6 0.4 187.1 634.1 70.50%

1991 58.6 0.5 215.8 692.6 68.8

1992 65.4 0.6 240.8 767 68.6

1993 70.1 0.8 269 834.3 67.8

(c)

1994 73.3 0.9 269.4 914.1 70.5

1995 78 1.6 280.7 1016.9 72.4

1996 82.6 3.2 318.2 1127.9 71.8

1997 91 4.8 373.5 1254.5 70.2

1998 97.4 7.8 424.2 1416.6 70.1

Contd…
UNIT 24: Forecasting

1999 99.2 17.7 465.8 1582.3 70.6 369

S
2000 102.4 27.4 449.3 1751 74.3 Notes
2001 86.6 74.4 477.2 1909.5 74.1
___________________

9 Graph is petrol additives manufacturer. Perform the analysis ___________________

specified below and briefly discuss your findings in terms of

E
___________________
the managerial if not national implications.
___________________
(a) Quarterly sales figures in thousands for car petrol
___________________
additives are given below
___________________

UP
i. Construct a seasonal index.
___________________
ii. Predict sales for the first and second quarters of 2001. ___________________
iii. Construct a confidence interval for the 2001 Q2 ___________________
forecast.
___________________
Q1 Q2 Q3 Q4
1998 2 4 3 5
1999 4 7 5 10
2000 9 10 10 12
E-
(b) While car petrol additives have had difficulties getting off
the ground, Graph successfully introduced car diesel
additives. For a modified exponential trend and predict
2001 Q3 sales.
1998 Q1 Q2 Q3 Q4 1999 Q1 Q2 Q3 Q4 2000 Q1 Q2 Q3 Q4
2 2 7 20 40 70 150 200 250 400 750 50
CC

(b) Given data referring to sales of three wheeler full

additives and two wheeler fuel additives, calculate the
correlation coefficient by (a) parametric and (b) non-
parametric methods and test the significance.
3-wheeler 20 40 60 30 20 70

2-wheeler 2 3 4 4 2 6

Further Readings

Books
(c)

R S Bhardwaj, Mathematics for Economics and Business, Excel

Books, New Delhi, 2005
D C Sanchethi and V K Kapoor, Business Mathematics
Sivayya and Sathya Rao, An Introduction to Business Mathematics
 Quantitative Techniques for Management Applications

370
Web Readings

S
Notes
http://www.math.ucla.edu
___________________
www.textbooksonline.tn.nic.in
___________________

___________________ www.mathbusiness.com

E
___________________

___________________

___________________

UP
___________________

___________________

___________________

___________________
E-
CC
(c)
UNIT 25: Case Studies

Unit 25
371

S
Notes

Case Studies
___________________

___________________

E
___________________
Objectives
___________________
After analyzing these cases, the student will have an appreciation of the
concept of topics studied in this Block. ___________________

___________________

UP
___________________
Case Study 1: Exxon Oil Marketing Ltd.
___________________
After three months on the job at Exxon Oil Marketing Ltd.; Nidhi
Batra—a recent MBA graduate— was called into the office of Mr ___________________
Sunil, Vice President and General Manager. Mr Sunil said that
___________________
Nidhi’s initial progress at the firm had been quite satisfactory and
that, because she had acquired certain quantitative and computer
skills at the University and on the job, she was to be put in charge
of a sales and productions study. The study itself was rather
complex, but the technical problems boiled down to the following.
E-
Sales Portion of the Study

In order to get a better feel for a certain target market, average

annual family income in that market had to be estimated. Before
a sample could be drawn and income figures obtained, Nidhi had
to determine the proper sample size. She set the following
criteria: (a) maximum sampling error not more than 10,000 above
CC

or below the true population mean, (b) confidence interval should

be 99%, and (c) standard deviation of the population, based on a
previous study, was known to be 2000. What was the sample size?

Another aspect of the sales portion of the study called for the
determination of a seasonal index for trend. Nidhi decided to use
the method of simple averages and obtained the following input
data for the firm’s major sales outlet in the target market.

Quarterly Sales in lacs

Year Q1 Q2 Q3 Q4
1981 10 8 14 16
(c)

1982 18 16 22 24
1983 20 30 27 35

Question:

What was the seasonal index adjusted for trend for each quarter?
Contd….
 Quantitative Techniques for Management Applications

372 Production Portion of the Study

S
Notes
Here two technical problems were encountered. The first
___________________ consisted of a production forecast for 2001 based on the known
estimating equating log yc =1.082 + 0.013x with origin at July 1,
___________________
1994, x in one year units and y in thousand-ton units. What was
___________________

E
the forecasted tonnage?
___________________
Finally, in order to study Astra’s training program effectiveness;
___________________ Nidhi obtained pertinent data for five workers and calculated a
coefficient of determination. What was r2?
___________________

UP
Worker A B C D E
___________________
Hours of Training 1 2 3 4 5
___________________
Units Produced 4 3 6 5 9
___________________
Mindful that Mr Sunil would want to have some idea about the
___________________
implications, that is, validity or managerial explanation, of the
four value sets calculated, Nidhi wrote a brief statement
concerning each.

Question
E-
What would you have advised her to cover in each statement for
the four value sets?
Source: MBCQ-721D_Quantitative Techniques for Management Applications_CCE_UPES
CC
(c)
UNIT 25: Case Studies

Case Study 2: Parametric Correlation and Non-parametric 373

S
Correlation Notes
We do not know precisely why formerly great nations like the ___________________
Italian city states, Spain, France and England in this order
___________________
declined in economic and political importance during the modern
era. Exact data is not available. It is safe to assume, however,

E
___________________
that fiscal mismanagement was a major contributing factor in
___________________
each case. Today we do have data like the ones for India. Get this
from any good source. Perhaps one day historians trained in ___________________
statistics will perform appropriate analysis and know precisely ___________________

UP
the reasons for this nation’s decline.
___________________
Given your own understanding of our economic and political
___________________
situation today, briefly explain your findings after solving the
problems below. ___________________

Questions ___________________

1. Calculate the parametric correlation between Receipts and

Outlays for the 1970s and test the significance.

2. Calculate the non-parametric correlation between Receipts

E-
and Outlays for the 1990s and test the significance.

3. Compare your answers for 1 and 2. What does remain in each

case and is it operationally meaningful
Source: MBCQ-721D_Quantitative Techniques for Management Applications_CCE_UPES
CC
(c)
 Quantitative Techniques for Management Applications

374

S
Notes

___________________

___________________

___________________

E
___________________

___________________

___________________

UP
___________________

___________________

___________________

___________________
E-
CC
(c)
Glossary

Glossary
375

S
Notes

___________________

A Sample Event: The basic possible outcome of an experiment, it ___________________

cannot be broken down into simpler outcomes.

E
___________________
Absorbing State: A state is said to be absorbing (trapping) state if ___________________
it does not leave the state.
___________________

Assignment Problem: An assignment problem is a special type of ___________________

UP
transportation problem.
___________________
Average: An average can be defined as a central value around
___________________
which other values of series tend to cluster.
___________________
Categorical Data: Categorical data are data that take a finite set
___________________
of values that can be either numeric or categorical.

Column Matrix: A matrix having only one column.

Commercial Averages: Commercial averages are the applications
of averages in commercial situations.
E-
Conditional Probability: The probability of an event occurring,
given that another event has occurred.
Continuous Probability Distributions: Continuous probability
distributions are those which represent continuously variable
random variables
CC

Correlation Coefficient: The degree of correlation between

normally distributed dependent and independent variables is
signified by the correlation coefficient.
Cycling Process: A cycling (periodic) process is one in which
transition matrix contains all zero elements in the retention cells
(diagonal elements) and all other elements are either 0 or 1.
Data Presentation: The presentation of complex mass of data in
a simple way so that it becomes easier to understand.
Data: The facts and figures that is collected, analyzed and
interpreted.
(c)

Decision Maker: An individual/group responsible for making a

choice of appropriate course of action.
 Quantitative Techniques for Management Applications

376
Decision Tree: A decision tree is a decision flow diagram that

S
Notes
includes branches leading to alternatives one can select among the
___________________
usual branches leading to events that depend on probabilities.
___________________
Determinant: A numeric value that indicate singularity or non-
___________________
singularity of a square matrix.

E
___________________
Deterministic Variables: Deterministic variables can be
___________________ measured with certainty.
___________________ Discrete Probability Distributions: Discrete probability

UP
___________________ distributions are those where only a finite number of outcomes are
possible.
___________________
Effective Demand: When a consumer is willing and able to
___________________
purchase some quantity of a commodity at the existing market
___________________
price, he is said to have an Effective Demand for that good.

EPPI (Expected Profit with Perfect Information): It is the

maximum obtainable expected monetary value based on perfect
information as to which state of nature will occur.
E-
Equation: A statement that two expressions (connected by the
sign=) are equal.
Event: Any set of all possible outcomes or simple events of an
experiment.

EVPI (Expected Value of Perfect Information): It is the

CC

maximum amount one would be willing to pay to obtain perfect

information.

Exponential Function: The function that associates the number

ex to each real value of x is called exponential function.
Frequency Distribution: A frequency distribution is the principal
tabular summary of either discrete or continuous data.

Function: A function f from a set X (domain) to a set Y, a subset of

X Y, is a rule that associates a unique element of set Y (target) to
each element in X.

Heuristic Variables: Heuristic variables are those that exist in

(c)

highly complex, unstructured, perhaps unknown decision making

situations.
Intervening Variables: Those variables that may affect the
dependent variable's magnitude but which are not considered in
the decision space of the forecast either because of oversight or
Glossary

their effect is deemed negligible, are known as intervening 377

S
variables. Notes

___________________
Linear Equation: A Linear expression equated to zero is called a
linear equation. ___________________

E
Linear: The relationship between the variables is directly ___________________

proportional. For example, if a wooden table requires 30 cubic feet ___________________

of wood then 10 tables would require 300 cubic feet wood.
___________________
Logarithmic Function: The function that associates login to x is ___________________

UP
called the logarithmic function.
___________________
Mathematical Averages: Mathematical averages are those which
___________________
utilize mathematical formula for the calculation of their values.
___________________
Matrix: An array of numbers arranged in certain numbers of rows
___________________
and columns.
Median: Median is the value of that item in a series which divides
the series into two equal parts, one part consisting of all values
less and the other all value greater than it.
E-
Mixed Strategy: When the player has alternative courses of
action and he has to select combination of these with some fixed
probabilities.
Mode: Mode is that value which has the maximum frequency (i.e.
occurs most often) in a given set of values.
Moving Average: The moving average is an arithmetic average of
CC

data over a period and is updated regularly by replacing the first

item in the average by the new item as it comes in.
Non-sampling Errors: Errors due to calculations or improper
convention of observation are called non-sampling errors.
Numerical Data: Numeric data are data that exist in numeric
form, such as height, the number of children in a household and
annual income.

Objective Function (Z): The linear function which is to be

optimized is called Objective Function.
(c)

Objective of Assignment Model: Objective of assignment model

is to assign a number of resources to an equal number of activities
Observation: The set of measurements obtained for a single
element in the data set.
 Quantitative Techniques for Management Applications

378
Optimal: It means if a programme maximizes or minimizes some

S
Notes
measure or criterion of effectiveness. Ex. Maximization of profit /
___________________
sales or minimization of cost or distance etc.
___________________
Payoff: It is the effectiveness of particular combination of a course
___________________
of action and state of nature. These are also called as conditional

E
___________________ values/profits.
___________________ Positional Averages: Positional averages do not use
mathematical calculations but give you an indication about the
___________________
positional characteristics of certain items.

UP
___________________
Probability Distribution: A probability distribution is a rule that
___________________
assigns a probability to every possible outcome of an experiment.
___________________
Probability: A numerical measures of the likelihood of occurrence
___________________ of an uncertain event.
Program: A program is a set of instructions arranged in a logical
sequence.
Pure Strategy: It is the decision rule which is always followed by
the player to select the particular course of action.
E-
Quadratic Equation: An equation of the form, where a 0, b, c are
constant, is called a quadratic equation.
Qualitative Data: Data that are labels or names used to identify
an attribute of each element. Qualitative data use the nominal or
ordinal scale of measurement and may be non-numeric or numeric.
CC

Quantitative Data: Data that indicate that how much or how

many of an element. Quantitative data use the interval or ratio
scale of measurement and are always numeric.
Random Variable: An event whose numerical value is
determined by the outcome of an experiment is called a variate or
often a random variable.

Rectangular Matrix: A matrix consisting of m rows and n

columns, where.

Root: A value of the variable which satisfies the given equation is

(c)

called a solution or root.

Row Matrix: A matrix having only one row.

Saddle Point: In a payoff matrix saddle point is one which is the
smallest value in its row and the largest value in the column.
Glossary

379
Sampling Design: A sampling design is a definite plan for

S
Notes
obtaining a sample from the sampling frame.
___________________
Sampling: Sampling can be defined as the selection of some part
of an aggregate or totality on the basis of which a judgment or ___________________
inference about the aggregate or totality is made.

E
___________________
Simple Random Sampling: A procedure of sampling will be ___________________
called simple random sampling where individual items (units)
___________________
constituting the samples are selected at random.
___________________

UP
Singular Matrix: A matrix whose determinant is zero.
___________________
Skewness: Skewness may be defined as the lack of symmetry or
___________________
degree of distortion from symmetry exhibited by a probability
distribution. ___________________

___________________
Square Matrix: If the number of rows of a matrix is equal to its
number of columns, the matrix is said to be a square matrix.
Standard Deviation: The standard deviation of a sample SD is
similar to the mean deviation in that it considers the deviation of
E-
each X value from the mean.
Statistics: The art and science of collecting, analyzing, presenting
and interpreting data.

Stochastic Variables: Stochastic variables are characterized by

uncertainty.
CC

Strategy: The strategy for a player is the list of all possible

courses of actions that he will take for every pay-off that might
arise.
Transient State: A state is said to be transient if it is not possible
to move to that state from any other state except itself.
Transportation Model: The transportation model of linear
programming is a flow optimization technique.
(c)
 Quantitative Techniques for Management Applications

380

S
Notes

___________________

___________________

___________________

E
___________________

___________________

___________________

UP
___________________

___________________

___________________

___________________
E-
CC
(c)