Capture-Recapture模型与车险事故总体预测  

Capture-Recapture Models and Prediction of Accident Population in Auto Insurance

作　　者：郭念国[1] GUO Nian-guo(College of Science,Henan University of Technology,Zhengzhou 450001,China)

机构地区：[1]河南工业大学理学院,河南郑州450001

出　　处：《数学的实践与认识》2021年第3期110-116,共7页Mathematics in Practice and Theory

基　　金：河南省高等学校重点科研项目(19A910001);2015年度中国保监会部级课题项目资助。

摘　　要：车险事故总体预测问题一直是车辆保险公司研究的重点内容之一,目前最为常用的方法是与泊松分布相关的模型.基于车辆保险中索赔数据的结构特征,构建了Capture-Recapture模型,并使用一组车辆保险数据,利用Capture-Recapture及常用的零膨胀泊松等模型分别建模分析,得出了一些新的结论,即Capture-Recapture模型拟合效果整体较优,从而为车辆保险公司更好预测事故总体提供一定的理论依据.The population estimation of accident in auto insurance is one of primary problems in the study of property insurance companies,and the Poisson distribution and associated with it may be most popular used.In this paper,based on the structure of claim data,we show some properties of Capture-Recapture models,and modeled different models and compared each other by using a real claim data of auto insurance.Some new results were obtained,namely Capture-Recapture models are overall superior to zero-inflated Poisson regression etc.,which provides a theoretical basis for the insurer to comprehend the information of the accident.

关 键 词：事故总体 Capture-Recapture模型 泊松分布 

分 类 号：F842.634[经济管理—保险]