错误先验假定下Bayes线性无偏估计的稳健性
The Robustness of Bayes Linear Unbiased Estimations under Misspecified Prior Assumption
-
摘要: 本文基于错误的先验假定获得了一般线性模型下可估函数的Bayes线性无偏估计(BLUE), 证明了在均方误差矩阵(MSEM)准则和后验Pitman Closeness (PPC)准则下BLUE相对于最小二乘估计(LSE)的优良性, 并导出了它们的相对效率的界, 从而获得BLUE的稳健性.Abstract: The Bayes linear unbiased estimator (BLUE) of parameters is derived for general linear model under misspecified prior assumption. The robustness of BLUE over ordinary least square estimator (LSE) are shown in terms of mean square error matrix criterion and Pitman closeness criterion, the bounds of the relative efficiency of estimators are obtained.