基于因子分析与K-means聚类耦合的分时保费定价方法研究

Research on Time-sharing Premium Pricing Method Based on Factor Analysis and K-means Clustering

传统车险定价基于"一人一车"前提,与租赁汽车所有权与使用权分离的运营模式冲突。为了解决租赁汽车分时保费的定价问题,研究了基于驾驶员行为的费率因子分级设计方法。首先,采用因子分析法提取驾驶行为因子,以方差贡献率的加权均值作为权重,设计了驾驶员综合风险评价因子S;然后,以S为聚类指标利用K-means聚类算法实现对驾驶员风险的自动分级,进而为分时保费费率因子的分级提供依据;最后,以某大型租赁公司所提供的19位驾驶员实车数据作为样本,结合某保险公司所提供的违章及出险事故数据,证明了该方法的可行性与有效性,该方法可为以人为中心的车险保费定价提供积极的理论参考。

The traditional auto insurance pricing is based on the premise of ＂one person-one car＂,which conflicts with the operation mode of the separation of ownership and use right of the leased car. To solve the pricing problem of the leased car＇ s time-sharing premium,it has studied the method of rate factor grading design based on driving behaviors. Firstly,it extracted the driving behavior factors by factor analysis and design the complex risk evaluation factor S of drivers based on the weighted mean of variance contribution. Then,the automatic classification of driver risk is realized by using the k-means clustering algorithm with S as the clustering index,it can provide the basis for the classification of premium rate factor. At last,according to the sample analysis of 19 drivers＇ real vehicle data provided by a large leasing company,the feasibility and validity of this method is proved by combining the data of the illegal and accident records provided by an insurance company. This approach provides a positive theoretical reference for the implementation of UBI.