Scholarlink

These days, machines are getting better at deciphering human behavior and its underlying meanings. Predicting customer churn is one of the many areas in which this capacity can be applied. Churn prediction identifies customers who are likely to leave a company's services, making it a valuable tool for businesses focused on improving customer retention. Customer retention is essential to banks’ long-term health and growth. By identifying key factors leading to churn, companies can develop loyalty programs and develop strategies to retain customers. The goal of this study is to build predictive models to identify customers at risk of churning, employing six popular machine learning algorithms: Logistic Regression (LR), Decision Tree (DT), Random Forest (RF), K-Nearest Neighbors (KNN), eXtreme Gradient Boosting (XGBoost), and Support Vector Machine (SVM). To enhance performance on imbalanced data, we employ Synthetic Minority Over-sampling Technique (SMOTE) coupled with Tomek Links (SMOTE-Tomek). The study examines the effect of SMOTE-Tomek on model performance across four well-known metrics: Accuracy, Precision, Recall, and F1-Score. Overall, the results show that RF and XGBoost is the top-performing models, with Random Forest delivering the most balanced performance across all evaluation metrics. Both models achieve a promising accuracy of 86%, demonstrating the potential of this approach in improving bank customer churn prediction.

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Enhancing Bank Customer Churn Prediction from an Imbalanced Dataset Using Machine Learning and SMOTE-Tomek Resampling

  • 会议论文

  • 发表日期：

    2026年

查看全文
  • Othman Berrada Chakour,
  • Abderrahim Maizate,
  • Abdelaziz Ettaoufik,
  • Khalid Aissaoui

摘要

These days, machines are getting better at deciphering human behavior and its underlying meanings. Predicting customer churn is one of the many areas in which this capacity can be applied. Churn prediction identifies customers who are likely to leave a company's services, making it a valuable tool for businesses focused on improving customer retention. Customer retention is essential to banks’ long-term health and growth. By identifying key factors leading to churn, companies can develop loyalty programs and develop strategies to retain customers. The goal of this study is to build predictive models to identify customers at risk of churning, employing six popular machine learning algorithms: Logistic Regression (LR), Decision Tree (DT), Random Forest (RF), K-Nearest Neighbors (KNN), eXtreme Gradient Boosting (XGBoost), and Support Vector Machine (SVM). To enhance performance on imbalanced data, we employ Synthetic Minority Over-sampling Technique (SMOTE) coupled with Tomek Links (SMOTE-Tomek). The study examines the effect of SMOTE-Tomek on model performance across four well-known metrics: Accuracy, Precision, Recall, and F1-Score. Overall, the results show that RF and XGBoost is the top-performing models, with Random Forest delivering the most balanced performance across all evaluation metrics. Both models achieve a promising accuracy of 86%, demonstrating the potential of this approach in improving bank customer churn prediction.