Cox模型多因子共线性处理方法的进一步研究

Further Study on Methods in Handling Multi-variable Collinearity in Cox Regression Model

目的 进一步探讨Cox模型中出现多因子共线性时的处理方法。方法 首先应用因子分析法将多因子的信息分解为公因子和特殊因子两部分 ,然后对这两部分信息分别进行Cox模型拟合。结果 实例分析对分解后的公因子进行Cox模型拟合 ,并指出对特殊因子拟合的方法。结论 本文提出的方法可以避免信息的损失 。

Objective To discuss and handle the problem of multi-variables collinearity in Cox regression model.Methods First,to divide the multi-variables into common factors and unique factors through factor analysis procedure.Second,to fit Cox regression model with common factors and unique factors respectively.Results To fit Cox regression model with divided common factors in the example,and point out how to fit Cox regression model with unique factors.Conclusion The method presented in this paper could avoid losing of information,and enable the models to be explained entirely.