Research On User Behavior Insurance Ratemaking Model Based On Driving Behavior Evaluation And Risk Probability Prediction

Vehicle insurance business has always occupied the vital position in China's property insurance business,but the overall profit situation of the auto insurance industry has been not ideal.Under the background of the deepening reform of commercial vehicle insurance rate and the continuous development of Internet of Vehicles technology,it is of great theoretical and practical significance to study the User Behavior Insurance ratemaking theory and method based on driving behavior data.Based on the theoretical research status of vehicle insurance ratemaking and the application status of User Behavior Insurance in domestic and foreign markets,this paper puts forward two UBI based ratemaking models from the perspective of vehicle insurance risk probability prediction and driving behavior scoring.The main contents of this paper are as follows:(1)Establish the risk probability prediction model.Based on the research progress of vehicle insurance ratemaking factors,this paper extracts 16 driving behavior factors from real user driving behavior data,establishes the risk probability prediction model by using logistic regression model and three classical machine learning models.Combined with AUC and F-measure,XGBoost model performs best in the study of risk probability prediction,and the experiment proves that the introduction of driving behavior factor can effectively evaluate the risk level of driver accident.(2)Establish the driving behavior scoring model.The index system of driving behavior scoring model is established by concerning the relationship between driving behavior factors and driving safety,the weight of which is determined according to the improved EW-AHP.The effectiveness of the scoring model is verified through correlation analysis.(3)Establish two UBI ratemaking models based on risk probability and driving behavior scoring model.According to the risk probability predicted by XGBoost model and the score given by driving behavior scoring model,the claim risk level of drivers is evaluated by clustering algorithm,and the rate adjustment coefficient is determined accordingly.Two UBI vehicle insurance ratemaking models based on claim risk levels are formed.Experiments show that the matching level between vehicle insurance pricing and driving risk will reach a new height as long as the driver behavior factor is involved in the rate determination of vehicle insurance.