|
Activity Number:
|
286
|
|
Type:
|
Contributed
|
|
Date/Time:
|
Tuesday, August 4, 2009 : 8:30 AM to 10:20 AM
|
|
Sponsor:
|
Section on Nonparametric Statistics
|
| Abstract - #305440 |
|
Title:
|
Estimation and gene selection in Microarray studies
|
|
Author(s):
|
William C. Bridges*+
|
|
Companies:
|
Clemson University
|
|
Address:
|
Department of Applied Economics and Statistics, Clemson, SC, 29634,
|
|
Keywords:
|
parametric ; non parametric ; dimension reduction ; gene selection
|
|
Abstract:
|
The support vector machine (SVM) and the penalized logistic regression (PLR) have been successfully applied to microarray cancer diagnosis problems. We propose the modified penalized logistic regression (MPLR) model to estimate a new conditional probability which is of interest in microarray studies, but has not been addressed in the literature. The method can also be used for generating initial order of genes in cancer microarray studies. Empirical results on breast cancer prognosis data indicate that the MPLR has the ability to select the important genes. Therefore, we believe our method is a great addition to the existing gene selection methods.
|
