线性回归模型复相关系数的影响评价

在线性模型中，复相关系数作为模型拟合程序的度量是一个常用的统计量，但它对异常值及影响点是很敏感的，本文用数据删除法及局部影响分析讨论复相关系数的影响点的识别问题，在各种情况下都得了了相应的诊断统计量，最后用一实例加以说明。

Assessing Local Injluence for Multiple Correlation Coefficients in Regression Model

Multiple correlation coefficients in linear regression model is a importantstatistic as a measure of model fltting,however it is very sensitive for the presence ofoutliers and influential observations.In this paper,we discuss the identification ofintluential observations for multiple correlation coefficients in linear regression model byusing case-deletion and local influence analysis. The relative diagnostics are derived underthe variouse perturbation schemes,and finally a example is used for illustration.