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

  • 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\widehat\beta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βi在线性估计类中均方损失下的可容许估计.

     

    Abstract: 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 \widehat\beta 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.

     

/

返回文章
返回