Research And Application Of The Prediction Model Of Telecom Customers’ Arrears Based On XGBoost

At present,many domestic social service enterprises(communication,power supply,gas,water supply,etc.),their charging methods are all post-paid.For these enterprises,how to reduce arrears,reduce costs,and improve efficiency,and at the same time,differentiated processing and careful reminders to ensure customer perception and avoid user loss are important issues that need to be solved urgently.To this end,it is necessary to predict potential arrears customers in advance and classify them.However,the existing arrears prediction models have shortcomings such as poor classification performance,low stability,and the generalization ability needs to be improved.The main work of this paper is to address the above shortcomings,using the telecom customer arrears data,research and build a telecom customer arrears prediction model based on XGBoost,and design and develop a telecom customer arrears prediction system based on SpringBoot and Vue framework.The main research work and achievements include:1.Deeply study and analyze the relief and RFE algorithms,and integrate them to form a second-order feature selection algorithm based on relief RFE.Integrate the advantages of the above two algorithms to generate the arrears feature set of telecom customers.2.A telecom customer arrears prediction model based on xgboost is proposed,and the parameters of the model are optimized based on specific application scenarios and requirements.Finally,the experiment is compared with the traditional classifier logical regression,decision tree and random forest.The experiment shows that the classification performance,stability and generalization ability of the model are improved compared with the traditional model.3.Based on the telecom customer arrears prediction model,a telecom customer arrears prediction system is developed by using springboot and Vue framework,which realizes the effective prediction and analysis of customer arrears.Main contributions: The research presents a series-structured Relief-RFEbased second-order feature selection algorithm to generate a telecom customer arrears feature set;optimizes the application of the XGBoost algorithm,and pr oposes a telecom customer arrears prediction model;Design and implement a t elecom customer arrears forecast management system.