混合系数线性模型h-K型岭估计的改进研究

Research on Improvement of h-K Class Ridge Estimator in Mixed-Effect Coefficient Linear Model

文章对病态的混合系数线性模型Z(t)=[X(t)]′α+[Y(t)]′β提出一类新的估计h-K型估计,讨论此种估计的相关性质,利用Stein式压缩技术证明了可以改进岭估计(在均方误差意义下);同时给出参数的最优值满足的条件,证明h-K型岭估计的可容许性.

The paper first proposes the h-K class of estimators of the coefficients of the mixed-effect linear model Z（t）=[X（t）]′α＋[Y（t）]′β in the ill condition.The ridge estimators is improved by applying the idea of the James-Stein regression on this class of biased estimators under the mean square error criterion;then the condition that the value of the parameter satisfies is given. This class of biased estimators is proved to be ad？missible.