In order to discover potential churn customers of banks in advance,a customer churn rate prediction model based on SMOTE and GBDT algorithm is constructed.The model first performs data preprocessing,and then constructs new features that may affect the churn rate by analyzing the relationship between various features in the data set.Next,the SMOTE algorithm is used to deal with the imbalance of data distribution.Finally,the data set is input into the integrated learning GBDT classifier for prediction.The SMOTE and GBDT algorithms proposed in this article have achieved improvements in AUC,G-mean,and F1 score indicators.
万方数据