Predictive Modeling for Customer Churn

Abstract

This thesis focuses on predicting customer churn in the telecommunications industry using machine learning techniques. The goal was to identify the main factors that drive customers to leave and provide data-based insights for improving retention strategies. Several models were developed and compared, including Logistic Regression, Support Vector Machine, Decision Tree, Random Forest, Extra Trees, and XGBoost. Visual and statistical evaluations confirmed that the models captured clear behavioral patterns behind churn. Overall, the study demonstrates how predictive analytics can support telecom companies in anticipating customer loss and designing more effective retention actions.