生长曲线模型的多元广义压缩最小二乘估计类

The Multivariate General Compression LS Estimate of Regression Coefficient in Growth Curve Model

本文提出了生长曲线模型回归系数阵B的一类有偏估计──多元广义压缩LSE类.以改善设计阵呈病态时的LSE，讨论了它的可容许性，优效性以及在均方误差意义和Pitman接近原则下，改善LSE的条件，另外还讨论了它的相合性、最优性，Bayes性等优良性。

In this paper, multivariate general compression LS estimate B_GS (W1,W2 )of theregression cofficient B is considered when the design matrix present ill-condition in growthcurne moelel. And the admissibility is discussed, We show the regions preferable to LSE undermean square error and pitman's clossness repectively and also discuss the ather quality of B_GS(W1, W2), forexample consistency, Bayes quality and so on.