半相依回归系统中回归系数估计的若干结果  

SOME RESULTS ON ESTIMATING THE REGRESSION PARAMETERS IN SURS

作　　者：丁树良[1] 

机构地区：[1]江西师范大学,南昌330027

出　　处：《应用概率统计》1994年第3期243-252,共10页Chinese Journal of Applied Probability and Statistics

摘　　要：本文首先给出了半相依回归系统中回归系数的一种新的估计方法,即用一些较简单的矩阵代替β的BLUE(?)中相当复杂的矩阵,得到β的相应的估计.协方差改进估计(CIE)可以看成是BLUE的某种“近似”.当CIE可以改进时,我们引入广义CIE(GCIE),讨论了GCIE的一些统计性质.在∑_(mxm)已知时.我们还得到β_(?)在线性估计类中均方损失下的可容许估计.To estimate the regression parameters in seemingly unrelated regression system (SURS) a new method is proposed in this paper, i.e. some complicated matrix in BLUE β of β replaced by a series of simpler matrices (or a simpler matrix) and the uubiased estimator of β is obtained respectively. The covariance improvement estimate (OIE) may be considered as an approximation of BLUE in a sense. When OIE can be improved, the genaralized OIE(GCIE) is introduced and some statistical properties of GOIE is discussed. When ∑>0 is known, the admissible linear estimates of βi is obtained.

关 键 词：半相依回归 回归系数 估计 CIE 

分 类 号：O212.1[理学—概率论与数理统计]