Customer Churn Prediction Using Machine Learning Approaches

Customer Churn (CC) is a significant problem that companies and large organizations alike must be concerned about. Due to the direct impact on revenue, telecom sectors, in particular, are working to enhance ways for anticipating potential customer turnover. This study examines the several machine learning algorithms that are employed to build the churn model, which aids telecom operators in forecasting which customers are most likely to leave. To choose the optimal model among different approaches, the experimental data are compared. As a result, in terms of F1-score, the Random Forest plus SMOTE-ENN combination performs better than everything else. Our investigation indicates that the greatest forecast based on F1-score is 95 percent. Keywords—Class Imbalance; Machine Learning; Customer Churn; Entity Extraction