Some connections between Bayesian and non-Bayesian methods for regression model selection

被引：3
|
作者：
Liang, FM [1 ]
机构：
[1] Natl Univ Singapore, Dept Stat & Appl Probabil, Singapore 117543, Singapore
来源：
STATISTICS & PROBABILITY LETTERS | 2002年 / 57卷 / 01期
关键词：
Bayes factor; FPE alpha; criterion; Kullback-Leibler distance; MAP; variable selection;
D O I：
10.1016/S0167-7152(02)00048-2
中图分类号：
O21 [概率论与数理统计]; C8 [统计学];
学科分类号：
020208 ; 070103 ; 0714 ;
摘要：
In this article, we study the connections between Bayesian methods and non-Bayesian methods for variable selection in multiple linear regression. We show that each of the non-Bayesian criteria, FPEalpha, AIC, C-p and adjusted R-2, has its Bayesian correspondence under an appropriate prior setting. The theoretical results are illustrated by numerical simulations. (C) 2002 Elsevier Science B.V. All rights reserved.
引用
收藏
页码：53 / 63
页数：11
相关论文
共 50 条
12345