广义线性模型_五_陈希孺.pdf

D I :10.13860/j.cnki.sltj.2003.03.014 56 　　　　　　　　　22 　3 　2003 5 　 　　:1002—1566(2003)03—0056—08 () 陈希孺 (, 　100039) 　:本讲座是广义线性模型这 个题目的一个比较系统的介绍。 主要分3 部分:建模、统计分 与模型选择和诊断。 写作时依据的主要参考资料是L.Fahrmeir 等人的《Multivariate Statistical Modeling Based on Generalized Linear Models》。 :广义线性模型;建模;统计分 ;模型选择和诊断 :212 :A Generalized Linear Model s CHEN Xi-ru (Graduate School of Chinese Academia of Science, Beijing 100039, China) Abstract:This set of articles gives an introduction to generalized linear models.They can be divided into three parts: Model building, Statistical inference and Model diagnostics.The presentation is mainly based on L.Fahrmeir et al. 《Multivariate Statistical Modeling Based on Generalized Linear Models》. Key Words:generalized linear models.;model building;statistical inference;model diagnostics 　 2.1(MLE) () (X i, Yi ), i=1, …n。 Yi c(y i )exp(yi θi -b (θi ))dμ(y i ), i =1, …n (2.1) θi X i , β。 n L = c (y )exp(y θ-b (θ)) ∏ i i i i i=1 β=β n, L 。βMLE。 n n logL = logc(yi )+ (y iθi -b (θi )) ∑ ∑ i=1 i =1 c(y )β, β, 。 i l (θ)=y θ-b (θ), i =1, …, n (2.2) i i i i i ■ n n 　　　l (β)= l (θ(β))= (y θ(β)-b (θ(β)))