散度偏大计数数据分布的贝叶斯估计

Bayesian Estimates of Distribution for Count Data with Overdispersion

  • 摘要: 本文针对索赔次数数据的特点, 讨论了两类可导致散度偏大特征数据的分布类型: 零点膨胀分布与膨胀参数分布, 并根据Bayes理论与MCMC方法, 利用WinBUGS对其进行建模和抽样\bd 经过比较,给出了实现分布拟合的途径, 最后通过两个数值例子加以展示.

     

    Abstract: This paper deals with two classes distribution of count data with overdispersion: Zero-inflated Distribution and Inflated-parameter Distribution, which are accordance with data of claims. We consider several model formulations of those distributions by using Bayesian theory and MCMC methods in WinBUGS. By comparison, a approach of modelling data is obtained and two illustrations with real data are provided.

     

/

返回文章
返回