摘要
在连续测量数据的情况下,针对模型的复共线性,给出了混合系数线性模型参数的一类新的有偏估计d*ω=(C'M-1C+I)-1(C'M-1Z+ωd*),证明了在一定条件下该估计优于GLS和岭估计,并讨论了该估计的可容许性,最后又给出了该估计的广义形式.
In the repeated-measures data model, in order to deal with the muhiconllinearity, a new class of estimator d＊ω=（C＇M-1C＋1）-1（C＇M-1Z＋ωD＊）for the parameters in mixed effect linear model is proposed. Under certain conditions, the new estimator is shown to be superior to the GLS estimators and Ridge estimators. Its admis- sibility is discussed, and a generalized form of the estimator is presented.
出处
《河南教育学院学报（自然科学版）》
2014年第2期28-31,共4页
Journal of Henan Institute of Education(Natural Science Edition)
关键词
有偏估计
Liu估计
广义最小二乘估计
均方误差
可容许性
biased estimate
Liu estimate
generalized least squares estimate
mean square error
admissibility
