相依回归系统的主成分改进估计

相依回归系统的主成分改进估计 THE IMPROVED PRINCIPAL COMPONENTS ESTIMATE IN SEEMINGLY UNRELATED REGRESSION SYSTEM 安军 An Jun 1 first-author 武汉大学数学系,武汉,430072 武汉大学数学系,武汉,430072 Department of Mathematics, Wuhan University Department of Mathematics, Wuhan University 对两个相依方程组成的回归系统,当设计矩阵呈病态时,提出了回归系数的一个主成分估计β 1 **（k）及其相应的两步估计β 1 **（k,T）,并在均方误差意义下研究了它们的优良性质。 For a seemingly unrelated regression system consisting of two equations, in this paper, both a kind of improved principal components estimate and its two-stage estimate are proposed and their optimal properties are respectively discussed in the sense of mean squared error （MSE）. 最小二乘估计 主成分估计 均方误差 least squares estimate （LSE） principal components estimate mean squared error （MSE） O212.1 1996-03-01 2021-04-01 3