Activity Number:
|
382
|
Type:
|
Contributed
|
Date/Time:
|
Thursday, August 15, 2002 : 8:30 AM to 10:20 AM
|
Sponsor:
|
Section on Nonparametric Statistics*
|
Abstract - #301365 |
Title:
|
Variable Selection in Generalized Additive Models
|
Author(s):
|
Xiaoming Wang*+ and K. Carriere+
|
Affiliation(s):
|
University of Alberta and University of Alberta
|
Address:
|
632 CAB, Edmonton, Alberta, T2G 2G1, Canada 622 Central Arcade, Edmonton, Alberta, T6G 2G1, Canada
|
Keywords:
|
Generalized additive model ; Kernel estimation ; Nonparametric regression ; Dimentionality reduction ; Variable selection
|
Abstract:
|
Additivity is commonly used in statistical modeling. Additive models have proven to be an extremely useful statistical tool in high dimensional data analysis. In generalized additive models, we often find that only a subset of the components is nonzero. Each of those nonzero components is a function of a particular predictor variable, called a significant variable. In this paper, based on the marginal estimation method introduced by Linton and Nielsen (1995), we develop a new method of searching significant variables in generalized additive models. Our method can also be extended to searching for significant interactions in generalized additive models. Numerical examples demonstrate the practical aspects and the comparative advantages of the proposed method over the other existing methods.
|
- The address information is for the authors that have a + after their name.
- Authors who are presenting talks have a * after their name.
Back to the full JSM 2002 program |