Machine Learning-Based Prediction of Telecom Customer Churn: Comparative Model Analysis

Abstract

In the increasingly competitive telecommunications market, customer churn has become a key challenge affecting the profitability and sustainable growth of enterprises. This study aims to conduct an empirical analysis to compare the performance of various machine learning models in predicting customer churn in the telecommunications industry. In this study, Python and related machine learning libraries were utilized to build a prediction process, and strict performance evaluations were conducted on three models: Logistic Regression, Random Forest, and Extreme Gradient Boost. The evaluation indicators include accuracy rate, precision rate, recall rate, F1 score, and the area under the receiver operating characteristic curve (ROC-AUC). The research results show that the optimized XGBoost model exhibits the best performance in all evaluation indicators, demonstrating its outstanding ability to handle such classification problems. In addition, by analyzing the feature importance of the XGBoost model, this study identified the key drivers influencing customer churn, among which contract type, customer online duration, and monthly fee are the most significant predictors. These findings not only provide telecom operators with high-precision churn warning tools but also offer data-driven decision support for them to formulate precise and effective customer retention strategies.