Predictive Risk Modeling for Credit Default Probability

Design a comprehensive machine learning pipeline that predicts credit default probability using historical financial data. Develop a model that integrates multiple risk factors including macroeconomic indicators, individual credit history, and transaction patterns. Implement feature engineering techniques to extract non-linear relationships, and create a model evaluation framework that includes ROC-AUC, precision-recall curves, and cross-validation strategies. Demonstrate how the model can be operationalized for real-time risk assessment in banking environments.