主成分改进的Logistic回归模型方法在流行病学分析中的应用  

Application of modified logistic regression model in the analysis of epidemiology

作　　者：裘炯良[1] 郑剑宁[1] 张扬[2] 

机构地区：[1]宁波出入境检验检疫局卫生检疫处,浙江宁波315012 [2]浙江大学医学院流行病学教研室,浙江杭州310006

出　　处：《中国热带医学》2005年第2期207-209,共3页China Tropical Medicine

基　　金：卫生部基金项目(No WKZ-2001-1-17)

摘　　要：目的　探讨在涉及多自变量的Logistic回归分析中变量间多重共线性的诊断和处理方法。　方法　应用主成分改进的Logistic回归方法进行多重共线性变量的诊断与处理。　结果　去除了回归模型中变量间的多重共线性影响,建立了较为理想的关系模型。　结论　在Logistic回归模型分析中应用上述方法进行多重共线性的诊断和处理是有效及可行的。Objective To explore the diagnosis and treatment of multivariable multicollinearity in the logistic regression analysis. Methods The data with multivariable multicollinearity were diagnosed and treated using the logistic regression model method improved by principal component analysis. Results The effect of multicollinearity among variables were eliminated in the regression model and an ideal mathematical model was constructed. Conclusion The new method is effective and feasible for diagnosis and treatment of multivariable multicollinearity in the logistic regression model analysis.

关 键 词：LOGISTIC回归模型 流行病学分析 诊断与处理 LOGISTIC回归分析 多重共线性 Logistic回归方法 改进 自变量 结论 主成分 

分 类 号：R195.1[医药卫生—卫生统计学] G633[医药卫生—卫生事业管理]