错误先验假定下多元回归系数Bayes估计的性质  

PROPERTIES FOR THE BAYES ESTIMATOR OF REGRESSION COEFFICIENTS IN MULTIVARIATE LINEAR MODEL UNDER MISSPECIFIED PRIOR ASSUMPTION

作　　者：彭向阳[1] 周占功[1] 

机构地区：[1]湘潭大学数学系,湖南湘潭411105

出　　处：《常德师范学院学报（自然科学版）》2003年第2期5-7,共3页Journal of Changde Teachers University

基　　金：湘潭大学科研基金 [0 2XZX16]

摘　　要：在错误指定的先验假定下研究了多元回归系数估计 (BE) ,并在矩阵损失下对其与最小二乘法估计 (LSE)进行了比较 。Under misspecified prior assumption the Bayes estimator(BE)of regression coefficients in multivariate linear model were studied.Based on the matrix loss function,it was compared with least square estimator(LSE).Under the PPC criterion,the superiority of BE over LSE was discussed.

关 键 词：多元线性模型 多元回归系数 BAYES估计 先验假定 矩阵损失 后验Pitman closeness准则 

分 类 号：O212.8[理学—概率论与数理统计]