考虑离群值的对数正态模型及其预测分布的蒙特卡洛法实现  

Logarithmic Normal Model Considering Outliers and Its Prediction Distribution by Monte Carlo Method

作　　者：闫春 解冰心 YAN Chun;XIE Bing-xin(College of Mathematics and System Sciences,Shandong University of Science and Technology,Qingdao 266590,China)

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

出　　处：《统计与信息论坛》2018年第9期17-22,共6页Journal of Statistics and Information

基　　金：国家自然科学基金项目<基于结构化大数据深度挖掘的非寿险保险公司经营风险模型研究>(61502280)

摘　　要：在非寿险业务中,对未决赔款准备金进行预测分布有着重要的意义,而流量三角形中离群值的存在,会影响未决赔款准备金预测的准确性。在流量三角形中引入离群值,运用基于正态分布的一元离群值检测的孤立点挖掘算法挖掘离群值,考虑不同位置下离群值的不同修正办法。将蒙特卡洛法应用于未决赔款准备金评估的对数正态模型中,通过数值算例加以实证分析,得到了未决赔款准备金的预测分布。In the non-life insurance business,it is of great significance to predict the distribution of outstanding claims.The presence of outliers in the flow triangle will affect the accuracy of the prediction of the outstanding claims reserve.The outliers are introduced into the traffic triangles.Outliers are extracted using an outlier detection algorithm based on the normal distribution and an outlier detection algorithm is proposed to consider outliers at different locations.The Monte Carlo method was applied to the lognormal model for the assessment of outstanding claims reserves.The numerical examples were used to demonstrate the predicted distribution of outstanding claims reserves.

关 键 词：孤立点挖掘 对数正态模型 预测分布 蒙特卡洛法 

分 类 号：F840.65[经济管理—保险] O211.3[理学—概率论与数理统计]