A Data-Driven Machine Learning Framework for Behavioral Customer Churn Prediction and Personalized Marketing Targeting

Customer churn is a major challenge for e-commerce businesses because losing customers can affect long-term revenue and customer relationships. Businesses may increase customer retention and create more successful marketing campaigns by identifying customers who are likely to churn. However, variations in customer purchasing behavior and class imbalance can make accurate churn identification difficult. This study uses the Brazilian E-Commerce Public Dataset from Olist to provide an ML-based method for customer churn prediction and segmentation. The proposed approach focuses on identifying meaningful customer behavioral patterns and distinguishing between churn and non-churn customers. A Random Forest classifier is trained using an 80:20 stratified train-test split and evaluated using accuracy, precision, recall, F1-score, confusion matrix, and ROC-AUC. The proposed model achieves 99.95% accuracy, 99.93% precision, 98.98% recall, 98.96% F1-score, and an AUC of 0.9998 on the test set. These results demonstrate the model's ability to accurately identify customer churn while maintaining a low classification error rate. The results show that ML may support data-driven customer segmentation, retention strategies, and targeted marketing in e-commerce settings by offering insightful information about customer behaviour.