End-to-end credit risk modeling framework covering Probability of Default (PD), Loss Given Default (LGD), Exposure at Default (EAD), credit scorecard development, and Expected Loss estimation using Python and Basel II risk management principles.
This project demonstrates the complete credit risk modeling lifecycle used in banking and consumer lending. The workflow includes data preparation, risk modeling, scorecard development, model validation, portfolio monitoring, and Expected Loss estimation.
Financial institutions need reliable methods to assess borrower risk, estimate potential losses, and support data-driven lending decisions while maintaining regulatory compliance.
- Build Probability of Default (PD) models
- Develop Loss Given Default (LGD) models
- Develop Exposure at Default (EAD) models
- Create credit scorecards
- Calculate Expected Loss (EL)
- Monitor model stability and performance
Consumer lending dataset containing borrower demographics, credit history, loan characteristics, repayment behavior, and default outcomes.
- Data cleaning
- Missing value treatment
- Feature engineering
- Weight of Evidence (WoE)
- Information Value (IV)
- Logistic Regression
- PD Modeling
- LGD Modeling
- EAD Modeling
- ROC-AUC
- Gini Coefficient
- KS Statistic
- Population Stability Index (PSI)
- Python
- Pandas
- NumPy
- Scikit-learn
- Matplotlib
- Seaborn
- Jupyter Notebook
- Credit Risk Modeling
- Basel II
- Probability of Default (PD)
- Loss Given Default (LGD)
- Exposure at Default (EAD)
- Expected Loss (EL)
- Credit Scorecards
- Weight of Evidence (WoE)
- Information Value (IV)
- Population Stability Index (PSI)
- Data preprocessing and feature engineering
- Exploratory data analysis
- WoE and IV analysis
- PD, LGD, and EAD model development
- Credit scorecard construction
- Model validation and monitoring
- Documentation and project organization
Successfully developed an end-to-end credit risk framework capable of estimating borrower default risk, potential losses, and overall portfolio exposure.
- XGBoost and LightGBM models
- MLflow experiment tracking
- FastAPI deployment
- Power BI reporting dashboards
- Cloud deployment
credit_risk_modelling_python/
│
├── data/
├── notebooks/
├── models/
├── reports/
├── README.md
└── requirements.txt
This project was completed as part of the 365 Data Science Credit Risk Modeling in Python course. All code implementation, analysis, model development, and repository documentation were completed by me for learning and portfolio purposes.
MIT License