Streaming Platform User Churn Prediction Model

Create an advanced churn prediction model for a streaming platform using XGBoost and scikit-learn. Develop a comprehensive feature engineering approach that incorporates user interaction data, content consumption patterns, subscription history, and demographic information. Implement a multi-stage machine learning pipeline that not only predicts churn probability but also provides actionable retention strategies. Include model interpretability features to understand key churn drivers.