主成分估计的应用探讨

Discussions On Application of Principal Components Estimates

本文首次把主成分估计的方法应用于动物性状的多元线性回归模型中复共性线之克服，结果表明由此法所获得的回归参数的估计不但在数值上较最小二乘估计合理，而且还在一定意义上使得被共线性扭曲了的变量间的关系得到恢复，从而提高了模型的预测精度。

In this articl, first application of principal components estimates to eliminating themulti- collinearity existing in a multivariable linear regression model amollg animal traits is discussed.Rsults indicated that the principal components estimates not only was more rationd innumber than least Square setimates,but also made recuperated the relations distorted by the serious multicollinearity. Nence,a linear model with a higher predicting accuracy was obtained.