Customer Churn Prediction in Bank Based on Different Machine Learning Models

With the rise of internet finance, the competition in the banking industry has become increasingly fierce. Preventing the loss of customers and retaining old customers has become an important concern of major banks. Firstly, according to an existing open dataset, this paper makes a descriptive statistical analysis of each feature, and uses Logistic, Random Forest and Scv models to predict customer churn, such as AUC curves, Mean Absolute Error (MAE), Root Mean Square Error (RMSE), and Mean Absolute Error (MSE). The better-performing models listed above are then chosen. Finally, Svm is selected as the best performance model, and according to descriptive statistics and feature importance, some suggestions are put forward for banks to retain customers.