混合系数线性模型参数的一类新估计

A New Class of Estimators for Coefficients in Mixed Effect Linear Model

在连续测量数据的情况下,针对模型的复共线性,该文给出了混合系数线性模型参数的一类新的有偏估计,称之为s-K估计,在一定条件下证明了这类估计分别优于最小二乘估计、Stein估计以及岭估计.

In the repeated-measures data model,in order to deal with the multicollinearity, a new class of estimator called s-K estimators for the parameters in mixed effect linear model are proposed.Under certain conditions,the new estimators are shown to be superior to the Stein estimators,Ridge estimators and least squared estimators,respectively.