聚集数据多元线性模型参数的多元广义聚集双参数广义岭估计的相对效率
The Relative Efficiency of Generalized Ridge Estimation for Multivariate Generalized Aggregation Doubleparameters of Multivariate Linear Model with Aggregated Data
余新宏 1郑剑平 1王礼霞1
作者信息
- 1. 合肥经济学院基础课教学部,安徽合肥 230011
- 折叠
摘要
复共线性聚集数据模型的岭估计过于强调参数的稳定性,而忽略了参数的无偏性.针对这个问题,采用施坦(Stein)压缩、广义岭(Liu)型估计等处理方法,提出聚集数据线性模型参数的多元广义聚集双参数的广义岭估计[(B)h(K,D)=(X'T'TX+hI)-1(X'T'TX+QDQ')(X'T'TX+QKQ')-1 X'T'TY],讨论多元广义聚集双参数改进估计的优良性,证明其两种相对效率的问题,最后推导出这两种相对效率的上界.
Abstract
The ridge estimation of the multicollinearity aggregated data model overly emphasized the stability of param-eters and ignored their unbiasedness.In response to this problem,based on Liu-type and Stein-type compression estimation method,generalized ridge estimation[(B)h(K,D)=(X'T'TX+hI)1(X'T'TX+QDQ')(X'T'TX+QKQ')-1 X'T'TY]of multivariate generalized aggregation doubleparameters for linear model parame-ters of aggregated data was proposed,the superiority of the improved estimation of multivariate generalized aggrega-ted dual parameters was discussed,and the two relative efficiencies were proved.Finally,the upper bounds of these two relative efficiencies were derived.
关键词
聚集数据/多元广义聚集/广义岭估计/相对效率Key words
aggregated date/multivariate generalized aggregation/generalized ridge estimation/relative efficiency引用本文复制引用
出版年
2024
NETL
NSTL
万方数据