|Table of Contents|

The Bayesian Estimation of the Multiple Linear Regression Model Based on the Normal-Wishart Prior Distribution(PDF)

《南京理工大学学报》(自然科学版)[ISSN:1005-9830/CN:32-1397/N]

Issue:
2002年04期
Page:
434-437
Research Field:
Publishing date:
2002-08-30

Info

Title:
The Bayesian Estimation of the Multiple Linear Regression Model Based on the Normal-Wishart Prior Distribution
Author(s):
ZhuHuiming HangYuqi
School of Economics and Management,NUST,Nanjing 210094
Keywords:
mult iple regression Bayesian est imat ion matricvariate t dist ribut ion
PACS:
O212.1
DOI:
-
Abstract:
A thorough study is made on the Bayes theory about the multiple linear regression model Y= Xβ+ ε, in w hich the rows of the random error mat rix Eare independently dist ributed, each w ith a m-dimension normal dist ribut ion N m( 0, Σ) ,Σ> 0. According to the decomposit ion of the sample‘. s likelihood function, the norma-l Wishart is proven to be the parameters. conjug ate prior dist ribut ion. On the basis of the g iven prior information, the posterior dist ribut ions of the coef ficient matrix Band precision matrix 2- 1are mat ricvariate t dist ribution and Wishart dist ribut ion respect ively .

References:

1 张尧庭, 方开泰. 多元统计分析引论. 北京: 科学出版社, 19971 134~ 138
2 言茂松. 贝叶斯风险决策工程. 北京: 清华大学出版社, 19891 106~ 108
3 Zeller A. An intr oduction to Bayesian inference in econometr ics. New York: John Wiley & SonInc, 19871 396
4 Tong T L. The multivariate normal distribution. New York: Spr ing er-Verlay, 19921202~ 219
5 Kotz M, 吴喜之. 现代贝叶斯统计学. 北京: 中国统计出版社, 20001 174

Memo

Memo:
-
Last Update: 2002-08-30