Customer Churn Prediction in the Broadband Service on Machine Learning

With the development of the telecom industry, the telecom industry is gradually saturated. In order to increase profits, telecom companies must reduce the churn of old customers. As an important part of telecom customers, the study of broadband customer churn prediction is helpful for enterprises to find the customer churn and make effective measures timely. In this study, two feature classification methods, RF-RFE and SVM-RFE, and four prediction classification algorithms, namely Logistic Regression, KNN, SVM model and XGBoost model were compared. GridCV is used to adjust the parameters of the method to achieve the optimal effect. The assessment found that the XGBoost model performed better than other models in predicting broadband customer churn. This provides a good reference for predicting the churn of broadband customers.