Advanced Customer Churn Prediction Model with ML Feature Engineering

Develop a comprehensive churn prediction model for a SaaS company using Python and scikit-learn. Create a pipeline that: 1) Performs multi-stage feature engineering on customer interaction logs, 2) Uses stratified cross-validation with time-based splits, 3) Implements both XGBoost and Random Forest models, 4) Calculates SHAP values to explain feature importance. Provide code that handles class imbalance and generates a detailed model performance report with precision/recall curves.