线性模型中两步估计与BLUE差别的度量  

TWO STEP ESTIMATE OF LINEAR MODEL AND MEASURE OF BLUE DIFFERENCE

作　　者：李平玉[1] 燕子宗[1] 

机构地区：[1]湖北荆州师范学院数学系,湖北荆州434104

出　　处：《数学杂志》2004年第1期67-72,共6页Journal of Mathematics

摘　　要：本文讨论了广义线性模型中均值向量向和回归系数两步估计与最佳线性无偏估计差别的度量方法 ,给出了均值向量 (回归系数 )两步估计 μ( β)相对于其最佳线性无偏估计μ (β )的相对精度P( μ｜μ )的界及 μ- μ ( β- β )的欧氏范数界 .并把文章结果应用到两阶段抽样回归模型 ,方差非齐次回归模型 ,半相依回归模型中 .In the paper we discuss the measure methods about the mean vectors of expand linear model and the two step estimate and the best estimation difference of the regression coefficient, obtain the bound of P(μ *) and the Euclid norm bounded of -μ *(-β *), which is the ralative accuracy of the best linear nonskew estimation μ *(β *) to the two step estimation (), and apply it to the two stage sample regression model, the variance nonhomogeneous regression model and the halfdependent regression model.

关 键 词：线性模型 两步估计 BLUE 差别 度量 回归系数 均值向量 

分 类 号：O212[理学—概率论与数理统计]