Customer Churn Intelligence — ML Prediction & Retention Analytics

Data Science & Machine Learning · Scikit-learn · SHAP · Streamlit · IBM Telco dataset (7,043 customers)

An end-to-end machine learning system that predicts customer churn on the IBM Telco Customer Churn dataset, explains why each customer is at risk, and turns the results into retention actions through an interactive Streamlit dashboard. Because missing a customer who is about to leave costs far more than a false alarm, the models are tuned for ROC-AUC, PR-AUC and churn recall, with an adjustable decision threshold.

View project / source code