针对两个相依方程组成的一类回归模型,提出回归系数的一种组合主成分改进估计方法,在均方误差意义下讨论了这种估计量的优良性质,研究了此估计与主成份改进估计的关系.
For a class of seemingly unrelated regression system consisting of two equations,an improved estimation of principal components is proposed and the optimal properties are discussed in the sense of mean squared error(MSE).The relationship between the proposed estimation and that of the combinatorial principal components is developed.