Prediksi Churn Pelanggan Telekomunikasi Menggunakan Metode Supervised Learning dengan Random Forest dan XGBoost

Customer churn is a major challenge in the telecommunications industry, resulting in revenue losses. Therefore, the ability to predict customers at risk of churn is crucial for preventative measures. This study developed and compared ensemble-based churn prediction models, namely Random Forest and XGBoost, using historical customer data covering demographics, service, and usage aspects, through pre-processing, training, and model evaluation stages. The results show that both models perform well, but XGBoost excels in AUC and F1-Score metrics, indicating better discriminatory ability and precision-recall balance. Feature importance analysis identified key churn factors, such as Monthly Charges and Tenure, which provide a basis for companies to design more focused and effective retention strategies.