|
Activity Number:
|
6
|
|
Type:
|
Invited
|
|
Date/Time:
|
Sunday, August 3, 2008 : 2:00 PM to 3:50 PM
|
|
Sponsor:
|
Section on Statistical Computing
|
| Abstract - #300082 |
|
Title:
|
Functional Regression in a Marketing Context
|
|
Author(s):
|
Gareth M. James*+
|
|
Companies:
|
University of Southern California
|
|
Address:
|
Bridge Hall, 401R, Los Angeles, CA, 90089-0809,
|
|
Keywords:
|
Functional Regression ; Penetration Data ; Functional Clustering ; Functional PCA ; Marketing Data
|
|
Abstract:
|
Since Bass's landmark paper of 1969 parametric approaches such as the Bass model have played an influential role in prediction methods within Marketing. Nonparametric techniques such as FDA are relatively unknown in the Marketing Literature. In this talk we apply a Functional Regression approach to penetration curves corresponding to 760 product-country combinations. Our aim is to predict future penetration levels, time of product takeoff as well as time and level of peak sales. We compare the nonparametric functional approach to the Bass model in addition to several other approaches and demonstrate that considerable improvements in prediction accuracy are possible. In the process of producing our Functional Regression model we also illustrate that Functional Principal Components and Functional Clustering methods provide useful insights into the behavior of the penetration curves.
|
