An end-to-end automated credit risk scoring and validation framework designed for Model Risk Management (MRM) and Operational Risk Oversight.
CreditLens is a lightweight, high-performance credit default risk assessment engine. Built specifically to handle portfolio stress-testing and Probability of Default (PD) calculations, it provides real-time model interpretability, statistical model validation, and an interactive executive intelligence portal.
- Automated Risk Scoring Engine: Predicts live borrower default probabilities based on credit limits, repayment histories, and statement balances.
- Model Validation Metrics: Computes out-of-sample performance metrics including ROC-AUC scores and detailed classification reports.
- Interactive MI Dashboard: Built via Streamlit to enable real-time risk parameter stress-testing.
- Lightweight Architecture: Optimized for low-footprint execution.
CreditLens/
β
βββ data/
β βββ credit_data.csv # Primary credit default client dataset
βββ models/
β βββ trained_model.pkl # Serialized machine learning model artifact
βββ app.py # Streamlit web application interface
βββ train.py # Model training and validation script
βββ requirements.txt # Project dependencies