Advanced Multivariate Modeling for

Marketing
An Overview
Presentation
By
Dr. P. K. Viswanathan
Professor(Analytics)
Great Lakes Institute of Management
Chennai, India

What is Marketing Analytics?

Marketing Analytics(MA) can be defined
as the broad use of data and quantitative
analysis for marketing decisions within
organizations.
Adapted from Thomas H Davenport

Why Advanced Multivariate

Modeling(AMM) for Marketing?
Three significant events triggered the current
meteoric growth in the use of marketing analytics
in decision making:
Event1:
Technological developments, Revolution of Internet
and social networks, data generated from mobile
phones and other electronic devices, produce large
amount of data from which insights will have to be
sifted.
The discovery of pattern and trends from these data
for organizations will pave the way for improving
profitability, understanding customer expectations,
and appropriately pricing their products so that they
can gain competitive advantage in the marketplace.
3

Why Advanced Multivariate Modeling

for Marketing?

Event 2:
Advances in enormous computing power to
effectively process and analyze massive
amounts of data
Sophisticated and faster algorithms for
solving problems
Data Visualization for Business Intelligence

Why Advanced Multivariate Modeling

for Marketing?

Event 3:
Large data storage capability
Parallel computing, and cloud
computing coupled with better
computer hardware have enabled
businesses to solve large scale
problems faster than ever before
without sacrificing
5

Google Trends Graph of Searches on

the term Analytics

The graph displays the search volume for the word analytics from
2004 to 2013 (projected) on a percentage basis from the peak.
The figure clearly illustrates the recent increase in interest in
analytics.

Big Data
Big data: A set of data that cannot be managed,
processed, or analyzed with traditional
software/algorithms within a reasonable amount of time.
Big data revolves around
Volume Velocity Variety Value Veracity

Walmart handles over one million purchase

transactions per hour.
Facebook processes more than 250 million picture
uploads per day.

The Spectrum of Business Analytics

AMM-Tools
Classification:
Classification techniques(Discriminant Analysis, Logistic
Regression, Neural Nets, and CART) help in segmenting the
customers into appropriate groups based on key
characteristics.
For example, using Discriminant Analysis, an organization
could easily segment the customers into Long Term
Customers, Medium Term Customers, and Brand Switchers.
Another application in this context is classifying customers
into Buyers and Non-Buyers.
Classification will help professionals understand the
customer behavior and position their products and brands
using appropriate strategies.

AMM-Tools
Pattern Recognition:
A picture is worth thousand words and it reveals hidden
pattern in the data that could be leveraged by retail
professionals. Pattern recognition techniques include
Histogram, Box Plot, Scatter Plot and other Visual Analytics.
For example, histogram drawn for income of a particular
class of customers may reveal a symmetrical bell curve
pattern or may be left or right skewed.
Relationship between age and expenditure could be
captured using a scatter plot.
Box Plot enables Analytics Professionals to sift outliers
(extreme points) apart from providing the distribution
pattern.

AMM-Tools
Association:
Association Analysis helps in determining which of the
items go together. Association rules include a set of
analytics that focuses on discovering relationships that
exist among specific objects.
In this context, market basket analysis refers to an
association rule that generates the probability for an
outcome.
For example, market basket analysis may lead to a finding
that if customers buy coffee, there is a 40% probability that
they also buy bread.
Association rules can be adapted by organizations to store
lay out, items bundling, discount and sales promotion
decisions, and cross selling among others.

AMM-Tools
Predictive Modeling:
Both customer segmentation as well as identifying and
targeting most profitable customers can be facilitated by
predictive models.
Regression can be used for predicting the amount of
expenditure on a particular product based on input variables
income, age, and gender.
Organizations can leverage on other advanced models that
comprise Logistic Regression, Discriminant Analysis, and Neural
Networks for predicting a target variable as well as classifying
and predicting into which group the consumer belongs to.
For example, these models can classify and predict buyers and
non-buyers, and defaulters and non-defaulters on credit card
loan.

AMM-Tools
Prescriptive Modeling:
Optimization Analytics and Decision Analysis are the
hallmarks of Prescriptive Analytics. They prescribe a
course of action in unequivocal terms.
Linear Programming(LP) and Goal Programming are
the optimization models that could be fruitfully used for
media planning, market research and consumer choice
decisions.
Competitive Pricing, and New Product Launch using
Decision Analysis are very popular in marketing
strategic decisions

AMM-Future Outlook
Big Data Analytics will be the prime mover for discovering
pattern and trends.
Mobile Technology will be a major driver of marketing
analytics that will be incorporated in strategy formulation.
Interactions between Social Media and Customers
Perceptions will be a significant factor in determining
contributions/ profit margins.
Technology and Software will facilitate decisions on real
time basis.
Organizations must as soon as possible become proficient
in Marketing Modeling and tools of Analytics to become
winners in the marketplace.

Decision Analysis-Case Study

The Marketing manager of a company producing
consumer durable wants to decide whether to
launch a new product or not. The sales quantity
of the new product has two states of nature
namely high sales or low sales. The pay off table
is as
follows (Figures
in $ Million).
States
Probabili
Launc Do not

of
Nature

ty

Launch

High
Sales
Low
Sales

0.4

0.6

-4

0
15

Decision Analysis-Case Study

A market research agency is willing to do a study
that will cost $0.3million.
There are two
possibilities. The agency may predict high sales
for the new product or it may predict low sales
for the new product. Past record of the agency in
terms of predicting ability is tabulated below:

States
of
Nature

Prior
Agency
Agency
Probabili Predicts High Predicts Low
ty
Sales given
Sales given
States of
States of
Nature
Nature

High
Sales

0.4

0.6

0.4
16

Managerial Questions

a) Should the product be launched based

on EMV criterion?

b) Should the marketing manager take the

help of the agency or not?

17

Solution to the Case Study

a) The product should not be launched. EMV
for launching = (5)(0.4)+(-4)(0.6) = $0.4ml. Since this is less than $ 0(EMV
optimum) that you get if product is not
launched, the product should not be
launched.

18

Solution to the Case Study

b) In order to solve this problem, we need to
work out the revised probabilities of the
states of nature from prior probabilities of the
states nature. This means we want to obtain
posterior probabilities of high sales and low
sales
given
agency
prediction.
This
statisticians call as inverse probability
approach based on Bayes theorem. You can
easily get this by constructing joint probability
table given in the next slide

19

Solution to the Case Study

Joint Probability Table
States of Agency Agency
Nature
Predicts Predicts
High
Low
Sales
Sales
High
0.24
Sales
Low
0.18
Sales
Marginal 0.42
Probabilit
y of

Marginal
Probabili
ty of
States of
Nature

0.16

0.40

0.42

0.60

0.58

1.00
20

Solution to the Case Study

Now the calculation of revised probabilities is very
simple. Probability of high sales given agency
predicts high sales = 0.24/0.42 = 0.57. Probability of
low sales given agency predicts high sales
=0.18/0.42 =0.43. Extending this logic leads to the
following revised probability table.

Revised Posterior Probability Table

Posterior
Probability
of

Given Agency
Predicts
High
Sales

Low
Sales

21

Solution to the Case

Study

22

Explanation on the
Tree

The manager has two options in the first stage of

decision namely go for agency or no agency. This
appears in the first square node.
If you dont go for agency, then you have two decision
options-launch the new product or do not launch the
new product. These two options are shown in the next
square node. If you launch, states of nature (demand)
may be high sales or low sales with prior probabilities
0.4 and 0.6 respectively. Monetary values corresponding
to these two states of nature are given as 5 and 4 at
the extreme end of the branch. These are in $million. If
you dont launch, you get $0 million. EMV is calculated
in the usual way for these two options. They are 0.4
and 0 respectively. So, EMV optimum = 0. This appears
in the square node corresponding to no agency.

23

Explanation on the
Tree

If you decide to go for agency, there are two states of

nature. Agency predicts high sales with a probability of
0.42 and low sales with a probability of 0.58. These
probabilities are from the revised posterior probability
table.

If agency predicts high sales, you have again got twodecision options-launch or no launch. Under each of
these options, the states of nature could be high sales or
low sales. However, the probabilities of these states of
nature are posterior probabilities. These are given in
brackets. Please refer again revised posterior probability .

24

Explanation on the
Tree

EMV values are worked out as before by multiplying

probabilities of states of nature with corresponding pay
offs and then summing them. EMVs are displayed in
circle and square nodes in the diagram.

Please also note that while working out EMV values,

cost of doing the study by agency ($0.3million) is taken
out to represent the correct net monetary value. Under
the scenario of taking agency help, if you choose not to
launch, you still incur cost of $0.3million(-0.3 EMV)

When you finally fold back the tree, you find going for
agency produces a positive EMV of $0.17 million. This
is better than getting $0 million of no agency. Hence go
for agency.

25

Final Conclusions for this

Case Study
EVSI = EPSI-EMV optimum = 0.17-0 =$0.17
million. This is the gain you have got by
using sample information.
Take the decision of going to the agency.
If agency predicts high sales, then launch
the new product (EMV $0.83 million)
If agency predicts low sales, then do not
launch the new product (EMV $-0.3 million)
Expected value of sample information
(EVSI) is $0.17 million.
26

Investment in Stocks- Case Study

Mr. Joseph, investment manager of Wizard Finance has
$120000 to invest in new IT stocks. Data on four chosen
stocks are given below:

Stock

Dividend

Growth % Price per

Share

1.50

30

35

0.75

25

45

2.00

28

40

0.50

35

55

Case Study Cont.

Mr. Joseph has been asked to maximize growth in $ value
excluding dividends of this portfolio of shares in the coming
year. It is Wizards policy to get at least $2500 as annual
dividend from this portfolio of shares. Wizard also wants to
play safe by stipulating the following conditions. Amount
invested in stock 4 cannot be more than 55% of amount
invested in stock 1. Further the company requires that at
least 15% of total investment be in stock 2.

Formulate and solve this case as a linear programming

problem and obtain the optimum solution-using Solver.

Problem Formulation for the Case

Let
Let
Let
Let

Identify the decision variables

X1
X2
X3
X4

=
=
=
=

$
$
$
$

amount
amount
amount
amount

invested
invested
invested
invested

in
in
in
in

Stock
Stock
Stock
Stock

1
2
3
4

Problem Formulation for the Case

Formulate the objective function

From the wording of the problem, it is clear that Wizard

would like to maximize the growth of this portfolio of
shares.

Maximize Z = 0.30X1+0.25X2+0.28X3+0.35X4

Problem Formulation for the Case

Formulate the constraints

The amount set aside for investment is $120000. Hence

the total amount invested in the four shares has to be less
than or equal to $120000.

X1+X2+X3+X4
constraint)

<

120000(Investment

amount

Problem Formulation for the Case

Formulate the constraints

The company specifies an annual dividend of at least $2500 for this

portfolio of shares. This is a bit tricky. Since we know dividends per share
as well as the price of each share. The annual dividend from this portfolio
is
=

1.5
0.75
2.0
0.5

X1
X2
X3
X4
35
45
40
55

This must be greater than or equal to $2500. Hence this constraint is

1.5
0.75
2.0
0.5
> =2500(Dividend
X1
X2 constraint).
X3
X4
35
45
40
55

Upon simplification this constraint becomes

0.43X1+0.017X2+0.050X3+0.009X4 > =2500(Dividend constraint)

Problem Formulation for the Case

Formulate the constraints

Amount invested in Stock 4 cannot be more than 55% of amount invested in
Stock 1. This constraint can be written as X4 < = 0.55X1. Simplifying this is
written as

X4-0.55X1< =0(Safety constraint)

The last constraint is at least 15% of total amount invested must be in Stock2.
This can be written as X2 > = 0.15(X1+X2+X3+X4).
Simplifying this expression, this constraint becomes
0.15X1+0.85X2-0.15X3-0.15X4 > = 0(Minimum amount in Stock 2 constraint)

Finally X1, X2, X3, X4 > = 0(Non-negativity)

Complete Formulation at a Glance

Maximize Z = 0.30X1+0.25X2+0.28X3+0.35X4
Subject to

X1+X2+X3+X4 < = 120000(Investment amount constraint)

X4-0.55X1< =0(Safety constraint)

0.43X1+0.017X2+0.050X3+0.009X4 > =2500(Dividend constraint)

0.15X1+0.85X2-0.15X3-0.15X4 > = 0(Minimum amount in Stock 2

constraint)

X1, X2, X3, X4 > = 0(Non-negativity)

Spreadsheet Displaying Model

formulation for the Case

Spreadsheet Displaying
Optimum Solution for the Case

Computer Solution from Solver

The optimum solution is to invest :

Stock 1= $65806
Stock 2= $18000
Stock 3= $0
Stock 4= $36194
The maximum growth = $36909. 68