|
Activity Number:
|
151
|
|
Type:
|
Contributed
|
|
Date/Time:
|
Monday, August 7, 2006 : 10:30 AM to 12:20 PM
|
|
Sponsor:
|
IMS
|
| Abstract - #305695 |
|
Title:
|
Variable Selection via Penalized Likehood in Semiparametric Regression
|
|
Author(s):
|
Xiao Ni*+ and Daowen Zhang and Hao Zhang
|
|
Companies:
|
North Carolina State University and sanofi-aventis and North Carolina State University
|
|
Address:
|
Department of Statistics, Raleigh, NC, 27695-8203,
|
|
Keywords:
|
penalized likelihood ; smoothly clipped absolute deviation ; smoothing spline ; linear mixed models ; partial linear models
|
|
Abstract:
|
We introduce an estimation and model-selection procedure for semiparametric regression models via doubly penalized likelihood. The smoothly clipped absolute deviation (SCAD) penalty is used for variable selection. The proposed method not only selects important variables and estimates their coefficients simultaneously, but estimates the unknown nonparametric function with smoothing splines. We propose a linear mixed model framework in which the unknown parameters are computed iteratively. This framework also enables us to estimate the smoothing parameter as a variance component directly. Theoretical properties of the semiparametric estimators are explored. Simulation results are presented to provide empirical support. Further extension of our proposed method to correlated data will be discussed.
|
