摘要
汽车保险索赔次数通常存在过离散问题,继续使用泊松回归模型可能会低估参数标准误差,高估其显著性水平。文章利用双泊松分布模型处理此类问题,并结合一组汽车保险实际数据进行拟合,改善了拟合效果。
There exists over-dispersed problem in automobile insurance claim counts.If we ignore the over-dispersion and apply the standard Poisson model,we will underestimate the standard errors and overestimate the significance of regression parameters.The paper discusses double Poisson model and applies the model to a set of actual automobile insurance claim data.The results show that double Poisson can improve the goodness-of-fit than Poisson when the data is over-dispersed.
出处
《河南机电高等专科学校学报》
CAS
2012年第1期30-32,61,共4页
Journal of Henan Mechanical and Electrical Engineering College
关键词
过离散
泊松回归
双泊松回归
索赔次数
over-dispersed
Poisson regression
Double Poisson regression model
claim counts