Customer Churn Prediction Using MachineLearning: A Comparative Analysis of ClassificationModels

Abstract

Customer churn prediction is an important problem in industries that rely on recurring customers, especially in telecommunications where losing customers directly impacts revenue. However, accurately identifying which customers are likely to leave remains a challenge due to the complexity of customer behavior and imbalanced data. In this study, multiple machine learning models were developed and compared using the Telco Customer Churn dataset, including Logistic Regression, K-Nearest Neighbors, Support Vector Machine, Random Forest, Gradient Boosting, and XGBoost. The models were evaluated using accuracy, precision, recall, F1-score, and ROC-AUC to ensure a well-rounded performance comparison. Among all models, Logistic Regression achieved the highest F1-score (0.6040) while maintaining strong overall performance, making it the most balanced model for this problem. These results demonstrate that simpler models can remain competitive against more complex approaches and that selecting the appropriate evaluation metric is critical when working with imbalanced datasets.