混合系数线性模型参数的几乎无偏LIU估计

Almost Unbiased LIU Estimator for Mixed-effect Coefficient Linear Model

该文研究了连续测试数据情况下的混合系数线性模型的参数估计问题.提出了混合系数线性模型参数的几乎无偏LIU估计,研究了参数估计的性质,并证明了在一定条件下此估计优于最小二乘估计、几乎无偏岭估计和LIU估计,最后还讨论了新估计的可容许性和有偏参数的选取方法.

This paper studies the parameter estimation problem of a mixed-effect coefficient linear model with continuous testing data. The almost unbiased LIU estimator(AULE) of the model is given, and the property of parameter estimation is analyzed.Under certain conditions, the new estimator is shown to be superior to the least squared estimator, LIU estimator and almost unbiased ridge estimator, respectively. At last, the admissibility of the estimator and bias parameter selection are discussed.