基于车联网数据的驾驶行为保险定价  

Driving behavior insurance pricing based on telematics data

作　　者：孟生旺[1,2] 黄一凡 Meng Shengwang;Huang Yifan(Center for Applied Statistics,Renmin University of China,Beijing 100872,China;School of Statistics,Renmin University of China,Beijing 100872,China)

机构地区：[1]中国人民大学应用统计科学研究中心,北京100872 [2]中国人民大学统计学院,北京100872

出　　处：《系统工程学报》2023年第1期121-130,共10页Journal of Systems Engineering

基　　金：教育部人文社会科学重点研究基地重大项目(22JJD910003);国家社会科学基金资助重点项目(22ATJ005)。

摘　　要：为了提高车险定价的准确性和公平性,从车联网大数据中提取广泛的驾驶行为风险变量,建立汽车保险的纯保费预测模型.采用Logistic回归和数据分箱方法构建驾驶行为风险因子,应用机器学习算法计算保单累积损失金额的预测值,确定合理的车险定价方案.在此基础上,对各个风险因子与车险纯保费之间的关系进行量化分析,最终得到完整的分类费率表.实证结果表明,本文的定价方法不仅提高了车险纯保费预测结果的准确性,增强保险公司的竞争能力,同时也满足保险定价的可解释性要求.车辆驾驶员可以根据划分的风险类别改善驾驶行为,有助于促进社会整体福利的提高.Aiming at improving the accuracy and fairness of automobile insurance pricing,this paper derived a wide range of driving behavior variables from telematics data and developed a predictive model for the pure premium of automobile insurance to improve the accuracy and fairness of automobile insurance pricing.Logistic regression and data-binning methods were used to construct driving behavior risk factors,and machine learning algorithms were applied to calculating the aggregate loss prediction,so as to determine the reasonable pricing framework of automobile insurance.On this basis,this paper quantitatively analyzes the relationship between each risk factor and the pure premium,and finally obtains a complete actuarial rating schedule.The empirical results show that this method not only improves the accuracy of pure premium prediction and the competitiveness of insurance company,but also meets the interpretability demand of insurance pricing.Vehicle drivers can improve their driving behavior according to the proposed risk categories,which is helpful to improve the overall social welfare.

关 键 词：驾驶行为保险 纯保费 损失预测 机器学习 

分 类 号：C812[社会学—统计学]