Research On UBI Rate Determination Method Based On Entropy Weight-Topsis And Clustering

With the development of big data technology and the popularization of Internet of Vehicles applications,China's auto insurance market is facing a new round of market-oriented reforms,and major insurance companies have the right to decide premiums,which brings huge opportunities and challenges for the company's development.However,the currently used auto insurance rate determination model and method have been difficult to meet the requirements of auto insurance pricing.It is worth noting that UBI has become an inevitable trend in the development of auto insurance,and its method of determining rates has also received widespread attention.At present,research on UBI is mainly carried out from two aspects of driving behavior indicators and comprehensive evaluation of driving behavior.The methods for comprehensive evaluation of driving behavior mainly include analytic hierarchy process,entropy weight-analytic hierarchy process,support vector machine,etc.Subjectivity,affecting·experimental results.This article applies the entropy weight-Topsis model and clustering method to determine the UBI rate.The main work is as follows:(1)Explained the research background,significance and research status at home and abroad.(2)Introduced the Lasso regression method,Logistic regression model and random forest model,and the comparative analysis method of selecting the optimal model;introduced the modeling steps of the entropy weight-Topsis model and the currently widely used cluster analysis method.(3)Using Lasso regression method to select driving behavior indicators,establish a risk probability prediction model-Logistic model and random forest model.Through comparison,it is found that the Logistic model has a higher accuracy rate for predicting the driver's risk,and the F value is the largest.The Logistic model is an optimal risk probability prediction model.At the same time,the predicted risk probability of 30 drivers in the test set can be obtained.(4)Construct an entropy weight-Topsis model to comprehensively evaluate the driving behavior of the driver.Based on the evaluation results,the K-means clustering method is used to determine the risk interval,and the predicted risk probability of some drivers is compared with the driving risk type.Analysis,and finally determine the rate adjustment coefficient of drivers at different levels of risk,and finally determine their rates.(5)Based on the research basis of this article,several suggestions are made.