Poisson回归模型的影响分析  

Explore Strong Influential Points and Example Analysis for Poisson Regression Model

作　　者：张登峰 张志明[2] 

机构地区：[1]陕西咸阳彩虹学校,陕西咸阳712046 [2]云南师范大学数学学院,云南昆明650500

出　　处：《广西民族大学学报（自然科学版）》2015年第1期56-61,共6页Journal of Guangxi Minzu University ：Natural Science Edition

基　　金：春晖计划(Z2009-1-65002)

摘　　要：文章对经典的Poisson回归模型,进行了参数估计和参数检验,并对模型进行了统计诊断,做出诊断图,判断出强影响点,并利用帽子矩阵分解原理及数理统计原理,探索挖掘出新的诊断图来判断强影响点,最后通过实证分析表明所研究的是有用和有效的.In this paper, based on introductions of the definition, parameter estimation of the classical Poisson regression model, this paper mainly researches on the Statistical diagnosis models by constructing the diagnosis statistics, and diagnosis index figure to look for strong influential points, and using the hat ma-trix decomposition principle and mathematical statistics principle to explore new effective diagnostic indicator diagram to determine strong influential points. At last, through the empirical analysis shows that the study is useful and effective.

关 键 词：Poisson回归模型 强影响点 帽子矩阵 诊断图 

分 类 号：O212.1[理学—概率论与数理统计]