摘要
本文详细阐述了解释变量的多重共线性在回归建模与分析中的危害作用，并指出目前常用的几种消除多重线性影响的方法，以及它们的不足之处。本文结合实证研究指出：利用一种新的建模思路—ＰＬＳ回归。
This paper sets forth the harmfulness of multi-correlation in regression modelling and analysis,and moreover,it indicates that the several mothods of eliminating multi-correlation which are common used are not quite efficient.With an example,it introduces PLS regression-a new regression modelling method,which can eliminate the bad effect that affect the accurity and the reliabilty of model on account of the high correlation of the argument variables.
出处
《数理统计与管理》
CSSCI
北大核心
1996年第6期48-52,共5页
Journal of Applied Statistics and Management
基金
国家自然科学基金
