混合系数线性模型参数的Stein估计

Stein Shrunken Estimates of Parameters for Mixed-Effect Coefficient Linear Model

在连续测量数据情况下，对混合系数线性模型给出了固定系数和随机系数的两种形式的Ｓｔｅｉｎ估计，并证明了在均方误差意义下，Ｓｔｅｉｎ估计要优于ＬＳ估计，最后还讨论了Ｓｔｅｉｎ压缩系数的选取方法．

In the repeated-measures data model, when the performance is linear with re- spect to the mixed coefficients which consist of fixed and random parameters, this paper proposes two kinds of Stein shrunken estimates of the parameters and discusses their prop- erties. At last, two methods for evaluating the shrunken values are introduced.