Comprehensive Evaluation Of Car-sharing Users And Vehicle Relocation Based On Empirical Data

With the improvement of people’s living standards,the current public transportation system can no longer meet the increasingly personalized and diversified travel demands.Therefore,as an emerging travel mode car sharing services gradually expand the market in the context of the sharing economy and become an important travel mode for people in modern society.In the operation and management of car-sharing enterprises,they mainly rely on vehicle relocation to meet the demands of more users,so as to improve their profits.However,in this process,enterprises often take users as a whole to make operational plan decisions,while ignoring the impact of user heterogeneity on the enterprise,which is not conducive to the long-term stable development of enterprises.Therefore,this paper takes car-sharing users as the research object,conducts comprehensive evaluation and prediction on users based on the analysis of user behavior rules,and establishes a vehicle relocation model based on the evaluation results,thus providing a certain theoretical basis for the operation of enterprises.Firstly,this paper conducts research on car-sharing users from two aspects: travel behavior and driving behavior,respectively.On the one hand,variables of user behavior are defined from the perspective of user value.Based on the expansion of the RFM model,the K-means clustering algorithm is used to cluster the user value into three categories:high-value users,low-value users and potential users.The behavioral patterns of users are analyzed from different travel aspects,the results show that different types of users have heterogeneous behavior characteristics.On the other hand,from the perspective of user driving safety,variables are defined for driving behavior,and the driving safety of users is clustered to obtain five types of users.The users’ driving behavior characteristics are analyzed and the results show that the users with the highest probability of rapid acceleration have the lowest driving safety.Secondly,a comprehensive evaluation indicator system for car-sharing users is established according to users’ value,travel stability and driving safety.Among them,the weight of the indicators is calculated by the entropy method and the analytic hierarchy process,and the optimal weight method is used to optimize the two kinds of weight,the optimized weight is consistent with the actual situation.On this basis,the Technique for Order Preference by Similarity to Ideal Solution(TOPSIS)method is used to comprehensively evaluate the user’s behavior,and the evaluation results show that highvalue users have higher evaluation scores.According to the evaluation result obtained by the TOPSIS method,the Gradient Boosting Decision Tree(GBDT)algorithm is used to learn and predict the comprehensive score of the new users.The result shows that the prediction accuracy is as high as 99%.Finally,on the basis of comprehensive evaluation of users,a vehicle relocation model based on users’ evaluation is constructed.Under the condition of maximizing enterprise profit,the impact of users’ evaluation on enterprise operating profit and carsharing system is contrastively analyzed.The results show that considering users’ evaluation for relocation can increase users’ travel times for the enterprise,improve the average demand satisfaction rate of the stations,and increase the enterprise’s operating profit and users’ average evaluation value.In addition,in the case of considering users’ evaluation,a reasonable allocation of the number of vehicles in the system can also increase the operating profit of the enterprise.