医用线性回归变量变换及诊断

Variable Transformation and Diagnostics for Linear Regression in Medicine

本文运用近代回归分析的原理和方法,将残差图和偏回归杠杆图作为诊断回归模型缺陷的工具,利用变量变换使资料满足线性回归模型的基本要求。实例分析表明,这些方法在诊断模型缺陷,决定相应处理措施以及改进模型回归效果等方面均取得了满意效果。

This article uses the techniques of modern regression to make the residual plot and partial regression leverage plot. Analyses of examples show that, it is very effective to use these met- hods for diagnoses of model inadequacy,for determination of a corresponding treatment and for improvement of regression