基于应变数线性变换的约束广义G-M估计(英文)  

Restricted General Gauss-Markoff Estimation Based on Linear Transformations of the Dependent Variable

作　　者：于义良[1] 

机构地区：[1]天津商学院基础课教学部,天津300134

出　　处：《天津商学院学报》2002年第3期5-7,14,共4页Journal of Tianjin University of Commerce

摘　　要：考虑线性变换对约束线性模型回归系数的影响问题 .证明了观测数据的线性变换对于约束模型的影响可以通过一个可用两步约束最小二乘法解决的约束回归问题进行分析 ,得到了回归系数的约束可估函数的约束最佳线性无偏估计不受变换影响的充要条件 .In this article we consider the influence of the regression coefficient the arbitrary rank restricted linear model based on linear transformations of the dependent variable. It is shown that the effect of the transformations may be analyzed through an associated restricted regression problem which is amenable to solution by two step restricted least squares. The necessary and sufficient condition in terms of arbitrary transformations is established for that there is a linear restricted estimable function of A′Y which is a restricted best linear unbiased estimation.

关 键 词：应变数线性变换 约束广义G-M估计 约束线性模型 两步约束最小二乘法 伪变量 回归系数 约束可估函数 

分 类 号：O212.1[理学—概率论与数理统计]