## MKTG 3509: Customer Data Analytics

21 Mar

Course Number:          MKTG 3509: Customer Data Analytics

Semester:                     Fall, 2017

Homework 4: Multiple Linear Regression Analysis

Objectives

The objectives of this exercise are to: 1) practice building and evaluating predictive models using multiple regression analysis, and 2) learn how to use a multiple regression prediction equation to score a database of prospective customers.

Background

Sara Beckman, director of marketing for the upscale specialty catalog, Tuscan Lifestyles, which specializes in cookware, tableware, linens, and decorative home accessories in the spirit of the Tuscany region of Italy, wants to acquire new customers and grow the customer base for Tuscan Lifestyles.

For several years, Tuscan Lifestyles, has maintained a loyalty program, whereby customers are assigned to one of three loyalty segments, Gold, Silver or Bronze, based on ‘Total Dollars Spent To Date’ with Tuscan Lifestyles. Sara’s team would like to identify prospective customers that have the potential to become ‘Gold’ in the future as these customers are more profitable than the other loyalty segments.

Over the past year, Sara has purchased external demographic and financial data on its customers and merged that data into the current marketing database. Sara would like to identify the characteristics (variables) that are predictive of ‘Total Dollars Spent To Date’, which is the variable that determines membership to the loyalty segments. She plans on using multiple regression analysis to identify a prediction equation that can then be used to score a newly acquired database of potential customers.

Scoring is a process where each prospective customer is given a score based on a multiple regression prediction equation. At that point only prospective ‘Gold’ customers that exceed a predetermined cutoff point will be selected for future communications from Tuscan Lifestyles.

Assignment

For this assignment, we will conduct in class all of the necessary work with SPSS. Using the output that we generated in class answer the questions on the following pages.

Enter all your answers into this document. Save the document and then upload the document to Blackboard.

Due Date:

Tuesday, November 7 at 8:00 p.m.

Question 1: Evaluate the following three Model Comparisons

Comparison 1:

 SSE(C) PC SSE(A) PA SSR R2 F Probability of F Statistic Interpretation of Findings (Include F, Probability of F and R2 ) Prediction Equation for Model A using Unstandardized Regression Coefficients

Comparison 2:

 SSE(C) PC SSE(A) PA SSR R2 F Probability of F Statistic Interpretation of Findings (Include F, Probability of F and R2 ) Prediction Equation for Model A using Unstandardized Regression Coefficients

Comparison 3:

 SSE(C) PC SSE(A) PA SSR R2 F Probability of F Statistic Interpretation of Findings (Include F, Probability of F and R2 ) Prediction Equation for Model A using Unstandardized Regression Coefficients

 Question 2:   Which set of variables gives the best prediction of the dependent variable? What is your evidence?

Question 3:

For the following three prospective customers in table below, use the prediction equation for the model which includes Owner Occupied Housing Value Index and Number of Credit Cards Owned to predict “Total Dollars in Spending” and loyalty group membership (Gold or Silver/Bronze)?

(Show your calculations below the table)

 Customer Owner Occupied Housing Value Index (oohvi) # credit cards owned (numcred) Predicted Total Dollars Spent (not Log10Dol) Predicted Loyalty Segment (Gold or Silver/Bronze) 1 101 1 2 75 1 3 136 2

Question 4:

If we must pay the list broker \$1.75 for every name that we have identified using our Linear Regression Model, what would be the total cost to acquire all the predicted Gold customers from the list of prospects?

Question 5:

What would be the total cost to acquire every name on the prospect list?

Question 6:

Quantify the savings achieved from using the Linear Regression Model versus purchasing the entire list of prospects.