|
Activity Number:
|
306
|
|
Type:
|
Topic Contributed
|
|
Date/Time:
|
Tuesday, August 5, 2008 : 2:00 PM to 3:50 PM
|
|
Sponsor:
|
Section on Statistics and Marketing
|
| Abstract - #300603 |
|
Title:
|
Modeling Customer Purchases and Lifetime Value in a Noncontractual Setting with Markov Models
|
|
Author(s):
|
Edward C. Malthouse*+ and Shyam Gopinath and Robert Blattberg
|
|
Companies:
|
Northwestern University and Northwestern University and Northwestern University
|
|
Address:
|
1870 Campus Drive, Evanston, IL, 60208,
|
|
Keywords:
|
markov chains ; purchasing behavior ; lifetime value ; customer relationship management
|
|
Abstract:
|
Understanding customer purchasing behavior is critical for the efficient allocation of marketing resources. This study focuses on the use of Markov models in estimating customer lifetime value (LTV). While others have proposed using first-order Markov Chains with states defined by recency and possibly frequency, the first-order assumption, small set of covariates, and method of estimating transition probabilities have not been questioned or studied. We present an empirical study of whether predictive accuracy is improved by (1) relaxing the first-order assumption and (2) adding covariates such as factors accounting for seasonality and trends, previous values of recency, transaction sizes as well as time-dependent covariates for future marketing activity. We use logistic regression models to estimate transition probabilities and traversing a binary tree to estimate expected future LTV.
|
