Poisson Log-normal回归模型的影响评价(英文)  

Influence Assessment for Poisson Log-normal Regression Models

作　　者：李泽慧[1] 赵为华[1] 谢晓方[2] 

机构地区：[1]江苏南通大学,江苏南通226019 [2]烟台海军航空工程学院,山东烟台264000

出　　处：《昆明理工大学学报（自然科学版）》2013年第4期102-108,共7页Journal of Kunming University of Science and Technology(Natural Science)

基　　金：Support by National Natural Science Funds of China(11171112)

摘　　要：本文利用EM算法研究了来自于Lognormal分布权重的混合Poisson模型,即Poisson Lognormal回归模型,从而利用基于完全数据似然函数的条件期望进行影响评价.基于数据删除模型和局部影响分析方法,分别得到了广义Cook距离、Q距离和三种不同扰动模型下的正则曲率度量等诊断统计量.In this paper, the EM algorithm is emptoyed as a basis to study the Poisson Log-normal models by treating them as a mixed Poisson regression with the weights from the Log-normal distributions. Diagnostic and lo- cal influence methods are then adopted for Poisson Log-normal regression models. Several diagnostic measure-ments such as generalized Cook distance, Q distance and normal curvature in three perturbation models are ob-tained respectively in both case-deletion model and local influence analysis based on the conditional expectation of the complete-data log-likelihood function in the EM algorithm.

关 键 词：POISSON Log-normal回归 EM算法 局部影响分析 数据删除 Gauss—Hermite积分 

分 类 号：TD952[矿业工程—选矿]