|
Activity Number:
|
162
|
|
Type:
|
Contributed
|
|
Date/Time:
|
Monday, August 3, 2009 : 10:30 AM to 12:20 PM
|
|
Sponsor:
|
Biometrics Section
|
| Abstract - #304144 |
|
Title:
|
Linear Model Selection for Nearly Replicated Data
|
|
Author(s):
|
Andrew Neath*+ and Zugui Zhang and Joseph Cavanaugh
|
|
Companies:
|
Southern Illinois University Edwardsville and The University of Iowa and The University of Iowa
|
|
Address:
|
Department of Mathematics and Statistics, Edwardsville, IL, 62026,
|
|
Keywords:
|
lack of fit test ; discrepancy function ; conceptual predictive statistic
|
|
Abstract:
|
The problem of selecting an appropriate model from a potentially large class of candidate models is an important statistical application. We define a linear model selection problem by the search for the input variables which minimize the error between the predicted response and observed response. Model selection criteria developed within this framework require an estimate of the true model variance. In this paper, we show how results from lack of fit tests based on near or exact replicates provide recourse for accurately estimating the true model variance.
|
