Credit Risk Predictive Model with Machine Learning

Develop a sophisticated credit risk prediction model using scikit-learn that integrates multiple data sources including historical loan performance, customer demographics, and macroeconomic indicators. Create a pipeline that handles feature engineering, handles class imbalance using SMOTE, implements cross-validation, and generates an interpretable risk scoring mechanism. The model should output probabilistic default risk with confidence intervals and support model explainability using SHAP values.