You are on page 1of 2

Discriminant Analysis (2 groups)

Case 1:

A nationalized bank has been in the Credit Card business for the last 16 years. During the past 24
months their Repayment Default has shot up considerably. Currently around 50% of existing cardholders
have been observed to have defaulted during this period. Even though the bank charges a penal interest
on all delayed payments, such a high default rate is putting a lot of pressure on the bank’s Recovery
mechanism and has now begun to impact the profitability of this division. The problem appears to lie in
the Screening mechanism used by the bank to evaluate Credit Card applicants at the point of card
allotment. Hence the bank desires to revamp its Screening mechanism.

For this purpose, they have conducted a suitable research. Initially a preliminary research was
conducted and the parameters that have a significant influence on Payment Default were identified.
These were:

a) Card Applicant’s Age


b) Monthly Household Income
c) Number of years married
Historical data was collected on the above variables from the Bank’s own records for the past 2 years and
existing Card holders were categorized into 2 groups as below:
a) High Risk (Code = 1)
b) Low Risk (Code = 2)

This was done based on the bank’s experience during the past 24 months.

A data extract of this research is provided below for your analysis.

Case Questions:

Q.1. Build a Discriminant model to categorize existing cardholders into High Risk and Low Risk categories.

Q.2. State the Classification Accuracy of this Discriminant Model.

Q.3. Determine whether the groups created by this Discriminant Function are distinctly different
statistically.

Q.4. Which Predictor variable is the best discriminator for Payment default?

Q.5. Identify a Discriminant Criterion which would enable the bank to classify future Card applicants into
High Risk and Low Risk categories based on their Discriminant Scores.

Justify your answers.

Case 2:
A National Retail Chain desires to identify a Discriminant Function that would enable the firm to
distinguish between Normal Customers (Code = 1) and Loyal Customers (Code = 2) based on the
following Customer characteristics:
a) Frequency of Purchases in a year.
b) Average Purchase amount.
c) Number of years purchasing.
Loyal Customers would be eligible for special benefits in order to retain them.
For this purpose, they have conducted a suitable research. A data extract of this research is provided
below for your analysis.

Case questions:

Q.1. Build a Discriminant model to categorize existing customers into Normal and Loyal categories.

Q.2. State the Classification Accuracy of this Discriminant Model.

Q.3. Determine whether the groups created by this Discriminant Function are distinctly different
statistically.

Q.4. Which Predictor variable is the best discriminator for Customer Loyalty?

Q.5. Identify a Discriminant Criterion which would enable the bank to classify Customers in future into
Normal and Loyal categories based on their Discriminant Scores.

Justify your answers.

You might also like