摘要
通过研究logistic回归模型的诊断,对模型的影响分析、均值漂移模型、异常点的score统计量进行了分析.推导出了判别强影响点或者异常点的诊断统计量,如广义cook距离、似然距离,讨论了局部影响分析,最后给出了实例分析.
This paper mainly focuses on the diagnostic methods for Logistic regression model. We Systematically studied the impact of the modle analysis ,the mean shift motile,Score statistic of outlier. At the same time , derived from the impact of the strong point of identification or diagnosis of abnormal point statistics such as the generalized Cook distance ,likelihood distance . And then discussed the local impact analysis ,and conducted a case study.
出处
《聊城大学学报（自然科学版）》
2011年第1期27-31,共5页
Journal of Liaocheng University:Natural Science Edition
基金
国家自然科学基金资助项目(11061036)
云南省教育厅自然科学基金资助项目(2010Y007)
