基于多重共线性的处理方法  

Based on Multiple Collinearity Processing Method

作　　者：满敬銮[1] 杨薇[1] 

机构地区：[1]中南大学数学科学与计算技术学院,长沙410075

出　　处：《数学理论与应用》2010年第2期105-109,共5页Mathematical Theory and Applications

摘　　要：多重共线性简称共线性是多元线性回归分析中一个重要问题。消除共线性的危害一直是回归分析的一个重点。目前处理严重共线性的常用方法有以下几种:岭回归、主成分回归、逐步回归、偏最小二乘法、Lasso回归等。本文就这几种方法进行比较分析,介绍它们的优缺点,通过实例分析以便于选择合适的方法处理共线性。Multicollinearity referred to as collinearity is a multi - linear regression analysis in a very difficult issue. How to eliminate the collinearity hazards regression analysis has been a priority. The literature at home and abroad to deal with serious collinearity methods commonly used are the following： Ridge regression, principal component regression, stepwise regression, partial least squares method, Lasso regression. In this paper, a comparative analysis of these methods and describe their advantages and disadvantages, easy to select the appropriate ways to deal with collinearity through the example analysis.

关 键 词：岭回归 主成分回归 逐步回归 偏最小二乘法 Lasso回归 

分 类 号：O212.1[理学—概率论与数理统计]