相依回归系统参数的广义岭型改进估计  

Generalized ridge improvement estimations for parameters of seemingly unrelated regression systems

作　　者：刘春梅[1] 

机构地区：[1]哈尔滨商业大学基础科学学院,哈尔滨150028

出　　处：《哈尔滨商业大学学报（自然科学版）》2012年第6期739-742,758,共5页Journal of Harbin University of Commerce：Natural Sciences Edition

基　　金：黑龙江省教育厅人文社科项目(12522056)

摘　　要：从设计阵的多重共线性角度出发,提出一种新的相依回归系统回归系数有偏估计——广义岭型改进估计.讨论了此估计的优良性,如:有偏性、可容许性等,在均方误差准则下证明了此估计要优于传统的岭型改进估计,并对同类估计进行了比较.In this paper,according to the approximate multi-collinearity of design matrix,a new based estimation of regression coefficient,i.e.generalized ridge improvement estimation,was proposed for seemingly unrelated regression systems.The properties,such as biased property,admissibility were discussed in detail.Moreover,it was proved that the proposed estimation was superior over the traditional ridge improvement estimation in the criterion of mean square error.Finally,the comparisons among covariance improvement estimators were discussed.

关 键 词：相依回归模型 广义岭型改进估计 均方误差 有偏估计 

分 类 号：O29[理学—应用数学]