CAJ | 학술논문

本文针对非寿险索赔次数回归拟合问题,以(a,b,0)零膨胀分布类为基础,简化其描述表达式,引入服从均匀分布的扰动量,将离散变量转化为连续变量,并通过Gaussian Copula实现边际分布的连接,给出模型的参数估计,通过对一组汽车保险索赔次数数据的实证分析和结果比较,表明采用Copula连接后的(a,b,0)零膨胀分布类回归模型有效地改善了拟合效果,并且避免了保险费率厘定时对索赔次数分布的选择。
본문침대비수험색배차수회귀의합문제,이(a,b,0)령팽창분포류위기출,간화기묘술표체식,인입복종균균분포적우동량,장리산변량전화위련속변량,병통과Gaussian Copula실현변제분포적련접,급출모형적삼수고계,통과대일조기차보험색배차수수거적실증분석화결과비교,표명채용Copula련접후적(a,b,0)령팽창분포류회귀모형유효지개선료의합효과,병차피면료보험비솔전정시대색배차수분포적선택。
According to the problem of non-life insurance claim frequency regression fitting, the paper takes(a,b,0) zero-inflated distribution class as a foundation, simplifies the description expression, introduces a jitters variable with uniform distribution, transforms discrete variables to continuous variables, and joint marginal distributions by Gaussian Copula function, puts forward the parameter estimation of distribution model. By the empirical analysis and results comparison, which using a group of auto insurance claims data, shows that:the(a,b,0)zero-inflated distribution class regression model jointed with Copula function is effective to improve the fitting effect, and avoids the choice of the distri-bution for claim frequency in rate making.