The Churn Compass: Navigating Customer Attrition Trends using Machine Learning

Customer attrition challenges businesses, leading to revenue loss and instability. This study introduces a machine learning framework for predicting churn using models like Random Forest, XGBoost, and LightGBM. Essential steps such as data preprocessing, feature selection, and validation enhance predictive accuracy. To address class imbalance, the Synthetic Minority Oversampling Technique (SMOTE) ensures a balanced dataset. Shapley Additive exPlanations (SHAP) values provide insights into key churn factors. Among tested models, LightGBM shows superior performance with a high Area Under the Curve (AUC) score. Additionally, a web-based application allows users to upload data, adjust churn thresholds, and visualize metrics like accuracy, precision, recall, and F1-score. The interactive interface includes visual tools such as a churn distribution chart and an ROC curve, helping users interpret results. This study highlights predictive analytics’ role in customer retention, equipping businesses with datadriven strategies to minimize churn. By integrating machine learning with a user-friendly web application, this research offers a scalable tool for businesses to anticipate behavior and take proactive measures.