线性回归系统两步协方差改进估计的优良性  

THE EFFICIENCY OF TWO STAGE COVARIANCE IMPROVED ESTIMATOR OF SEEMINGLY UNRELATED REGRESSION EQUATIONS SYSTEM

作　　者：刘金山[1] 严利清[2] 

机构地区：[1]五邑大学数理系 [2]中国科学技术大学

出　　处：《数理统计与应用概率》1997年第4期322-326,共5页

摘　　要：对于ｍ个相依回归方程组成的线性回归系统，本文研究了两步协方差改进估计在均方误差阵意义下的优良性，即在随机误差服从正态分布的假设下，当样本量充分大时，两步协方差改进估计与协方差改进估计一样好．For the system of Seemingly Unrelated Regression (SUR) equations given by y i=X iβ i+e i(i=1,2,…,m) ,we consider the efficiency of two stage covariance improved estimator of β i ,with respect to the mean squared error matrix criterion.Under normal distribution assumption on the random error,the two stage covariance improved estimator may perfect well as the covariance improved estimator intruduced by Wang Song qui ,for large sample size.

关 键 词：相依回归模型 估计 线性回归 协方差 优良性 

分 类 号：O212.1[理学—概率论与数理统计]