数据挖掘下车险续保概率及画像建模研究  

Research on probability and image modeling of continued paymentof vehicle insurance premium under data mining

作　　者：秦雅琴[1] 梁璐 许庚 周燕宁 QIN Yaqin;LIANG Lu;XU Geng;ZHOU Yanning(School of Transportation Engineering,Kunming University of Science and Technology,Kunming 650504,China)

机构地区：[1]昆明理工大学交通工程学院,云南昆明650504

出　　处：《重庆理工大学学报(社会科学)》2025年第3期145-156,共12页Journal of Chongqing University of Technology(Social Science)

基　　金：国家自然科学基金项目“山区公路复杂交通环境下交通安全设施效用评价方法”(71861016);科技部国家重点研发计划项目“车路协同系统要素耦合激励与协同优化方法”(2018YFB1600505)。

摘　　要：以我国车辆保险为主要研究对象,为续保概率模型及车险用户精准画像建模。通过主成分分析法提取8个主要成分,运用逐步回归模型对车险续保率与多个影响因素间的关系进行实证优化研究,建立续保概率模型,并利用支持向量机二元分类模型进行预测。为研究车险用户的画像建模,运用二阶聚类算法做出车险用户的精准画像。结果显示:建立的续保概率模型,准确率高达92.8%;车险续保用户具体分为3类。利用所建立的模型和统计结果,从影响客户决策的因素进行判断分析,并从财险企业的管理经营角度提出相应改善建议和优惠政策,以提高客户的续保概率。为车险用户精准画像可起到风险管控的作用,减少交通事故,管理社会冲突,维护社会秩序。Taking China’s vehicle insurance as the main research object,the research aims to establish a probabilistic model for renewal of insurance and an accurate portrait modeling of auto insurance users.Eight principal components were extracted by principal component analysis,and a stepwise regression model was used to conduct an empirical optimization study on the relationship between the auto insurance renewal rate and multiple influencing factors.A probabilistic renewal model was established,and a support vector machine binary classification model was used for prediction.In order to study the portrait modeling of auto insurance users,a second-order clustering algorithm is used to make accurate portraits of auto insurance users.The results show that the established renewal probability model has an accuracy rate of 92.8%and auto insurance renewal users are divided into 3 categories.The research uses the established model and statistical results to make a judgment and analysis from the factors that affect customer decision-making,and puts forward corresponding improvement suggestions and preferential policies from the perspective of the management and operation of property and casualty companies to increase the probability of customer renewal.Accurate portraits of auto insurance users can give full play to risk management and control,reduce traffic accidents,stabilize social conflicts,and maintain social order.

关 键 词：主成分分析 逐步回归 PCA-SVM 二阶聚类 画像建模 

分 类 号：U-9[交通运输工程]