多重共线性的产生原因及其诊断处理

随着计算机应用的普及 ,多元回归分析在生产、科研等实践中得到广泛应用。但是 ,在使用线性回归模型时 ,容易忽视应用条件 ,即自变量间不存在近似线性关系 ,否则会使结果不准确 ,甚至严重偏离变量间本来的依存关系。因此 ,有必要分析多重共线性的产生原因及对线性回归模型的影响 ,进而给出诊断多重共线性的常用方法 ,最后总结几种克服共线性的方法 ,并以卫生管理中的一个实例加以说明。

Cause of multi-collinearity and its diagnosis and treatment

In the applcation of multiple regression analysis,the application condition that there is no linear relationship between independent variables is apt to be overlooked, therefore the obtained result may become incorrect and even be far from the original relationship between the variables. In this paper,the cause of multi collinearity and its effect on linear regression are analyzed, and some useful ways of diagnosing and treating multi collinearity are introduced.An example of health management is presented to show the diagnosing and treating methods.