基于EM算法的Poisson-Gamma回归的影响诊断  

Influence Analysis of Poisson-Gamma Regression Models Based on EM Algorithm

作　　者：李爱萍[1] 解锋昌[1] 

机构地区：[1]南京农业大学数学系,南京210095

出　　处：《三峡大学学报（自然科学版）》2006年第1期83-86,共4页Journal of China Three Gorges University：Natural Sciences

基　　金：国家自然科学基金项目(10371016);南京农业大学青年科技创新基金项目(KJ04020)

摘　　要：对于Poisson-Gamma回归模型,将来自于Gamma分布的权重看作缺失数据;在此基础上,引入EM算法;从而利用基于完全数据似然函数的条件期望进行影响诊断分析,并且进一步基于正则曲率方法系统研究了各种扰动模型下的局部影响分析,得到相应的诊断统计量.最后,通过一个实例说明了所得统计量的有效性.For the Poisson-Gamma regression models, the weights from the Gamma distribution are regarded as missing data. Based on this fact, the diagnostic analysis of the Poisson-Gamma regression models is developed based on the conditional expectation of the complete-data log-likelihood function in the EM algorithm. The local influence based on the several perturbation schemes is discussed. Finally, a numerical example is given to illustrate the results.

关 键 词：Poisson-Gamma回归 局部影响 EM算法 正则曲率 扰动 

分 类 号：O212.2[理学—概率论与数理统计]