岭型主成分估计的优良性质  

Some Properties of the Combining Ridge and Principal Components Estimate

作　　者：李兵[1] 陈国华[1] 段复建[2] 

机构地区：[1]湖南人文科技学院数学与应用数学系,湖南娄底417000 [2]桂林电子科技大学数学与计算科学学院,广西桂林541004

出　　处：《桂林电子科技大学学报》2009年第2期128-130,共3页Journal of Guilin University of Electronic Technology

基　　金：湖南省教育厅资助科研项目(07C389)

摘　　要：用线性回归模型的一种有偏岭型主成分估计,证明岭型主成分估计在M SE和GM SE准则下优于最小二乘估计,而且比主成分估计更有效,在协方差阵准则下优于最小二乘估计。并且进一步得到了在均方误差意义下岭型主成分估计是可容许估计。We consider the combining ridge and principal components estimate （CRPCE） of the linear regression model. It was proved that The CRPCE is better than least squares estimate under the MSE and GMSE principal and more effective than the estimated principal component. It is also better than LSE under the variance matrix. We proved the CRPCE was an admissible estimate.

关 键 词：有偏估计 容许性 均方误差 

分 类 号：O212.1[理学—概率论与数理统计]