|
Activity Number:
|
425
|
|
Type:
|
Contributed
|
|
Date/Time:
|
Wednesday, August 1, 2007 : 10:30 AM to 12:20 PM
|
|
Sponsor:
|
Section on Statistics and Marketing
|
| Abstract - #310307 |
|
Title:
|
On Models for Identifying Likely Lapsers
|
|
Author(s):
|
Enping Mai*+ and Raja Velu
|
|
Companies:
|
Syracuse University and Syracuse University
|
|
Address:
|
311 Whitman School of Management, Syracuse, NY, 13244,
|
|
Keywords:
|
Lapse Models ; Inference ; Random effects
|
|
Abstract:
|
Identifying likely lapsers is very important to almost all firms. The two models that are proposed in the marketing literature to predict customer lifetime duration are the Pareto/NBD model by Schmittlein et al (1987) and the BG/NBD model by Fader et al (2005). They mainly differ how the dropout rates are modeled. While those papers consider only estimation of the models, in our research we extend their models. In addition we develop formal inference procedures, further incorporating both time varying covariates such as promotions, and variables that are fixed over time, such as demographic information. The results show that these can vastly improve the performance of the models.
|
