一类索赔次数的回归模型及其在风险分级中的应用

A Regression Model Based on Double Parameters Poisson Distribution and Its Applications to Risk Classification

本文引出了一类索赔次数的分布并研究了基于此分布的回归模型,这类分布包含两个参数,它可以视为Poisson分布的一种推广,而且相对于Poisson分布来说具有散度偏大(over-dispersion)的性质;基于这类模型,本文研究了两个经典实例展示其在风险分级中的应用.

A distribution about claim numbers has been introduced and the regression based on the distribution has been investigated in this paper. This distribution including double parameters, which is over-dispersion about Poisson, can be applied on risk classification. Statistics inference can be used to analyze the departure from Poisson model by testing the parameter of over-dispersion. Two classical examples on risk classification are studied again, one is about ship damage numbers and another is about car claims.