基于广义线性模型的主成分估计及实例分析

Principal Component Estimation Based on Generalized Linear Models and its Analysis of the Practical Example

在广义线性模型已有的参数估计的基础上对自变量参数中可能出现的复共线性关系时的参数估计进行了改进,应用主成分分析在线性模型中的方法将主成分估计应用到广义线性模型中去,并在均方误差下证明了主成分估计优于最大似然估计,并列举实际例子进行分析,说明了主成分估计的优越性.

In this paper, on the basis of the existing parameter estimation in Generalized linear model, the improvement of the parameter estimation is given when the independent variable parameter have the multicollinearity relationships, principal component estimation is applied to the generalized linear model using the method of principal component analysis in the linear model, and under the mean square error, the principal component estimation is proved to be better than maximum likelihood estimation, at last a practical example is listed to illustrate the advantages of principal component estimation.