Abstract:Aiming at the overdispersion problem of vehicle insurance claims data, based on numerical simulation method, this paper used Poisson model(Poisson), Negative Binomial model (NB), Generalized Poisson model(GP) to fit the vehicle insurance claims data with different degrees of overdispersion. Moreover, we compared the superiority of the three models with mean square error(MSE), average bias, AIC criterion, and BIC criterion. Finally, we got the superiority of three models and gave suggestions for selecting model in different conditions. |