克服复共线性的一种新方法

A NEW CLASS OF WAY TO COMBAT MULTICOLLINEARITY

本文提出一种新的有偏估计_w=(X′X+I)^(-1)(X′Y+w_(LS)),其中0<w<1为参数。理论和计算机模拟结果表明_w与岭估计_R有一样好的统计性质。

This paper proposes a new biased cstimate β_w=(X＇X+I)^(-1)(X＇Y+wβ_(LS)), Where 0<w <1 is a parameter, Theory and computer simulation results show that β_w has the similar good statistical property as β_R.