岭回归参数选择的主成分方法

The Choice of Ridge Regression Parameter from Principal Component Viewpoint

当线性回归模型中自变量出现共线时,最小二乘估计不再是良好估计。岭回归估计和主成分估计是新提出的两种估计方法,本文讨论了在某些条件下,这两种估计是很接近的,同时提供了一种选择岭回归参数K的方法,称为主成分选择法。文章还列举了两个典型例子来说明。

The least square estimator is no longer the good estinator, if some or all of explanatory variables are collinear in linear regression model. The ridge estimator and principal component estimator are the two new methods that are supplied. In this paper we discuss the ridge estimator can be made very close to the principal component estimator in Euclidean distance, in the mean time, it has provided a method to select ridge regression parameter K. Lastly, the two canonical examples are given.