Customer Behavioral Modeling & Retention Decision Support System

An end-to-end customer churn prediction and retention decision-support system designed for non-contractual service environments where churn is inferred from customer inactivity and behavioral patterns.

Project Details:
Developed a temporal behavioral modeling pipeline using customer activity, booking history, transactional behavior, engagement signals, and recency-frequency-monetary features.

Engineered rolling-window features to capture customer behavior over time and support early churn-risk identification in a non-contractual business setting.

Trained and evaluated machine learning and deep learning models, including XGBoost and LSTM, for customer churn prediction, retention modeling, and risk classification.

Applied SHAP-based model interpretability to identify key churn drivers and explain model predictions to non-technical stakeholders.

Designed a decision-support layer that translated churn-risk predictions into actionable retention recommendations for targeted customer intervention.