未决赔款准备金评估Clark方法及R实现  

Clark Method for Outstanding Claims Reserve Estimation and Its R Implementation

作　　者：闫春 刘倩 刘伟 

机构地区：[1]山东科技大学数学与系统科学学院,山东青岛266590 [2]山东科技大学计算机科学与工程学院,山东青岛266590

出　　处：《统计与决策》2018年第4期73-77,共5页Statistics & Decision

基　　金：国家自然科学基金资助项目(61502280)

摘　　要：文章在研究Clark建模思路的基础上,将Bootstrap法引入,运用重复抽样对模型的预测误差进行度量,得到了模型的预测分布和分位数等其他统计特征,并应用R软件对Clark方法进行算例分析。最后,将Boot-strap法下得到的预测误差与原方法的预测误差计算进行理论与实证的比较,分析了各自的特点与优势。Based on the research of Clark＇s modeling idea, this paper brings in Bootstrap method and uses repeated sampling to measure the prediction error of the model and obtain its forecast distribution, quantile and other statistical features. And then the paper employs R software to conduct an example analysis on Clark method. Finally the paper calculates the prediction error obtained under Bootstrap method and the original prediction error to make a comparison theoretically and practically and also analyzes their characteristics and advantages.

关 键 词：未决赔款准备金 极大似然估计 预测均方误差 BOOTSTRAP法 

分 类 号：F840[经济管理—保险]