J4

• 数学 • 上一篇    下一篇

二次损失函数下Poisson回归模型中参数的近似Bayes估计

赵志文1,2, 王德辉1, 李 涵1   

  1. 1. 吉林大学 数学研究所, 长春 130012; 2. 吉林师范大学 数学学院, 吉林 四平 136000
  • 收稿日期:2008-01-10 修回日期:1900-01-01 出版日期:2008-09-26 发布日期:2008-09-26
  • 通讯作者: 王德辉

Approximate Bayesian Estimation for Parameters inPoisson Regression under Quadratic Loss Function

ZHAO Zhiwen1,2, WANG Dehui1, LI Han1   

  1. 1. Institute of Mathematics, Jilin University, Changchun 130012, China;2. College of Mathematics, Jilin Normal University, Siping 136000, Jilin Province, China
  • Received:2008-01-10 Revised:1900-01-01 Online:2008-09-26 Published:2008-09-26
  • Contact: WANG Dehui

摘要: 针对Poisson回归模型中当给定参数的先验分布后, 其后验分布的具体形式较难得到的问题, 利用Γ分布的大样本性质, 给出了似然函数的近似形式, 并得到近似的后验分布. 模拟结果表明, 利用这种近似的后验分布进行统计推断, 得到的参数Bayes估计具有较好的稳定性及精度.

关键词: 二次损失函数, Poisson回归, 近似Bayes估计, 参数, 先验分布, 后验分布, 似然函数

Abstract: It is, more often than not, difficult to gain the concret posterior distribution of Poisson regression after priori distribution of parameters is given. This paper provides an approximate likelihood function of the parameters according to the property of the large sample of Γdistribution and their approximate posterior distribution. The simulation study shows that the Bayesian estimation performs well based on the approximate posterior distribution.

Key words: quadratic loss funtion, Poisson regression, approximate Bayes estimation, parameter, prioridistribution, posterior distribution, likelihood function

中图分类号: 

  • O212.7