CHAPTER THREE

REGRESSION, FORECASTING & NETWORK ANALYSIS

Regression Analysis:
•A statistical technique which is used to measures the
relationship between two or more variables.
•When the value of one variable is related to the value of
another, they are said to be correlated
•Correlation means an inter-relationship or association. For
instance, there is likely to be some correlation between the
heights and weights of people
•The variables may be:
–Perfectly correlated (move in perfect unison)
–Partly correlated (some inter-relation but not exact)
–Uncorrelated (no relationship between their movements)

•Using the graph, movement in one variable may cause

movement in the same direction. This is known as positive
correlation, e.g. the height and weight.
• Where movement in one variable causes a
change in the opposite direction in the
other variable, this is known as negative
correlation.
• The degree of correlation between two
variables can be measured as by the
formulae below:

• (r) being the coefficient of correlation give

an indication of the strength of the linear
relationship between two variables.
 x y xy
15 60 225 3600 900
24 45 576 2025 1080
25 50 625 2500 1250
30 35 900 1225 1050
35 42 1225 1764 1470
40 46 1600 2116 1840
45 28 2025 784 1260
65 20 4225 400 1300
70 22 4900 484 1540
75 15 5625 225 1125
424 363 21926 15123 12815
• Using the formulae given, solve
Relationships between Variables
• There are many occasions in business when changes in
one factor affect the movements in one or several other
factors
• For instance, a marketing manager may observe that
sales increase when there has been a change in
advertising expenditure.
• Movement in one factor or variable could be as a result
of combined movement in several other factors or
variables
• The responding variable is called the Dependent
• Whiles the predicting variable is called the Independent
Types of Regression Analysis
• Regression can be simple or multiple.
• Simple Regression:
• Where the dependent variable (y), is caused by one independent
factor
• That is where there is a single cause to the occurrence of an
activity, that is a simple regression.
• A simple regression equation is represented by
• Multiple Regression:
• Where the dependent variable (y), are caused by several
independent factors and not just one as in the basic model
• For instance, changes in demand for a product may depend the
price of the product, the price of substitutes, the level of incomes,
the consumer tastes and so on
• A multiple regression equation is represented by
• The purpose of this equation is to forecast or predict and to
determine how reliable your forecast is
• Calculating the values of (a) and (b)
• General equation is
• Formulae for (a):

• Formulae for (b):

• Scatter Diagram:
• When there is a causal or scattered relationship
between two variables. (horizontal or x-axis being
the independent or causing variable whilst the
vertical axis is used for the dependent variable

 
••
•••
• • •
•
•
 Characteristics of Simple Linear
Regression
• Simple linear regression model assumes that the relationship between the
dependent variable (y) and the independent variable (x) can be approximated
by a straight line.
• It is a useful means of forecasting when data has generally linear relationship
• To have confidence in the regression relationship calculated, it is preferable to
have a large number of observations
• Any form of extrapolation, including that which is based on regression analysis,
must be done with great care, since the observe values relationship and
conditions may change drastically.
• In some instances, a particular (dependent) value may depend on two or more
factors in which case, multiple regression analysis is employed.
• E.g. a firm may produced the following multiple regression equation:
– Overheads = 10,500+6.8x+9.2y+2.7z
– Where:
– x = Labor hours worked
– Y = machine hours
– Z = tonnage produced
 Questions
• The following data have been collected regarding sales
and advertising expenditure.
» Sales (¢m) Advertising (¢‘000)
» 8.5 210
» 9.2 250
» 7.9 290
» 8.6 330
» 9.4 370
» 10.1 410

• Plot the above data on a scatter diagram and, using

judgment, decide whether there is a correlation between
sales and advertising expenditure
• 2) Calculate r for the data in (1) and
interpret
• 3) calculate for the data in (1) and
interpret
• Find the value of a and b in the formula
from the following data relating to the cost
incurred at various output levels. (Use the
calculator and the direct formulae for a and
b as given)
• Output Level (Units) Cost incurred (¢)
» 40 812
» 55 890
» 68 955
» 73 948
» 82 1050
» 89 1100
» 94 1160
» 95 1095
» 103 1250
» 110 1380
 Importance of Regression
analysis in Business
• Used to predict Wage of employees per years of
experience or years of education or gender
• Current stock price depends on its past values
• How current sale depends on advertising of
competitors
• If we can estimate these relationship, then we
can best understand how the word businesses
operates and predict the future outcome
 Network Analysis
• It is a techniques developed to aid
management to plan and control projects.
• It shows inter-relationship of the various
jobs or tasks which make up the overall
project and clearly identify the critical
paths of the project.
• It can provide planning and control
information on the time, cost and resource
aspect of the project.
• A variety of names are used fot this
techniques and some commonly used are:
• Critical Path planning CPP
• Critical Path Analysis CPA
• Critical Path scheduling CPS
• Critical Path Methods CPM
• Programme Evaluation and Review Technique
PERT.
 Basic Network Terminology
• Activity
• This is a task or job of work which takes time and resources
eg build a wall.
• An activity is represented in a net work by an arrow as

• The head of the arrow indicates where the task ends and the
tail where the task begin.
• Event
• This is a point in time and indicates the start or finish of an
activity, or activities eg wall built, foundation dug.
• The event is represented in a network by a circle or node as
• Dummy activity
• This is an activity which does not consume time or
resource
• It is used to show clear, logical dependencies
between activities so as not to violate the rules for
drawing networks
• It is represented by a dotted arrow as ........
 Forecasting-Time Series
Analysis
• Attempting to predict the future
• Virtually every form of decision making
and planning activity in business involves
forecasting. Typical applications include:
• Production planning
• Inventory control
• Investment cash flows
• Cost projections
• Demand forecasts
• Advertising planning
• Budgeting etc
Types of Forecasting Techniques
•Quantitative Techniques
•These are techniques of varying levels of statistical
complexity which are based on analyzing past data
of the items to be forecast, e.g. sales figures, store
issues, costs incurred.
•However sophisticated the technique used, there is
the underlying assumption that past patterns will
provide some guidance to the future
•Qualitative Techniques
•These are techniques which are used when data
are scare, e.g. the first introduction of a new product.
The techniques use human judgments and
experience to turn qualitative information into
quantitative estimates.
•Time scale for forecasting lengthens
•Even when past data are available, so that
standard quantitative techniques can be
used, longer term forecast require judgment,
intuition, experience, flair etc that is
qualitative factors, to make them more useful.
•As the scale lengthen, past pattern becomes
less and less meaningful
•Examples of qualitative methods includes
Delphi Method, Market Research, Historical
Analogy.
 Time series analysis
•As the name implies, time series analysis
uses some form of mathematical or statistical
analysis on past data arranged in a time series
e.g. sales by month for the last ten years.
•Advantages
•Relatively simpler
•Disadvantages
•Past data may not be representative e.g. may contain
the results of recession or boom
•Less appropriate over longer term due to among
others, external pressures.
•In the face of fluctuation or unstable conditions, time
series may give poor results
•Time series analysis-moving average
Past Forecasts
sales produce
Months Actual Sales

Jan 450
Feb 440
Mar 460
Apr 410 450
May 380 437
Jun 400 417
Jul 370 397 423
Agu 360 363 410
Sep 410 377 397
Oct 450 380 388
Nov 470 407 395
dec 490 443 410
Jan 460 470 425 424
• The three monthly averages forecast were
prepared as:

• =450+440+460
3
• =450