Machine Learning-Based Customer Churn Prediction in E-Commerce Platforms Using Behavioral Data Analytics

Abstract

Customer churn prediction has become a critical challenge for e-commerce organizations as retaining existing customers is more cost-effective than acquiring new ones. This study proposes a machine learning-based customer churn prediction framework using behavioral data analytics to identify customers at risk of leaving an e-commerce platform. The proposed approach uses the Kaggle E-Commerce Customer Churn dataset and incorporates comprehensive data preprocessing, including missing-value handling, categorical feature encoding, feature scaling, and class balancing via the Synthetic Minority Over-sampling Technique (SMOTE). Recursive Feature Elimination (RFE) is employed to select the most relevant features, while K-Means clustering is used for customer segmentation. A Light Gradient Boosting Machine (LightGBM) classifier is then developed to predict customer churn, and Shapley Additive explanations (SHAP) are applied to improve model interpretability by explaining feature contributions. Experimental results demonstrate that the proposed model achieves 96.00% accuracy, 95.00% precision, 92.00% recall, 93.00% F1-score, and a ROC-AUC score of 95.1%, outperforming several existing approaches. The proposed framework provides an accurate, efficient, and explainable solution to support proactive customer retention strategies and improve long-term business profitability.