OPTIMIZING CUSTOMER RETENTION IN TELECOM: A CHURN PREDICTION APPROACH

Abstract

Customer churn is a major challenge in the telecom industry, impacting revenue and business growth. This project focuses on predicting churn by analyzing customer behavior and service usage patterns using the Telco Customer Churn dataset. Key attributes like contract type, payment method, monthly charges, and tenure are examined to identify factors influencing churn. Processing missing values, encoding categorical variables, and normalising numerical features prepares ML models like LightGBM and XGBoost for training. Model performance is ensured by F1-score, precision, recall,and Accuracy; feature selection defines the most critical elements. The best model is used by a Flask web app to anticipate churn in real time. This tool allows telecom companies to make data-driven decisions, proactively address customer retention issues, and improve business stability by minimizing churn rates.