逆χ~2分布下正态模型参数的后验分布及其抽样  

Some Posterior Distributions of Parameters in Normal Linear Model and Their Sampling Method under Couter χ~2 Distribution

作　　者：陈浩明[1] 陈应保[1] 李开灿[2] 

机构地区：[1]华中师范大学数学与统计学院,武汉430079 [2]湖北师范学院数学系,黄石435002

出　　处：《北京印刷学院学报》2008年第2期57-61,共5页Journal of Beijing Institute of Graphic Communication

基　　金：湖北省优秀创新团队资助项目(D200622001)

摘　　要：当正态线性模型中的参数取逆χ2先验分布时,得出了回归系数和误差方差的联合后验分布和边际后验分布,特别导出了回归系数二次型的分布。利用回归系数和误差方差联合后验分布的分解形式,给出了一种抽样方案。When the parameters in normal linear model have counter X^2 distribution, this paper gives the joint posterior distributions of the regression coffecient and error variance, and gives their marginal posterior distributions. Specially, the posterior distributions of a quadric form about the regression coefficient in the linear model is obtained. According to the joint posterior distributions of the regression coefficient and error variance, we propose a sampling method for the joint posterior distributions.

关 键 词：线性模型 无信息先验分布 真实先验分布 逆χ^2分布 回归系数 抽样 

分 类 号：O212.4[理学—概率论与数理统计]