Credit Risk Assessment & Loan Default Prediction system

This project builds an end-to-end Credit Risk Assessment & Loan Default Prediction System that estimates the Probability of Default (PD) for loan applicants. Using the Home Credit dataset, the system processes applicant demographics, financial data, and credit history through a machine learning pipeline. Multiple models were evaluated, and a calibrated model was selected to ensure reliable probability outputs. The system converts PD into actionable decisions (Approve, Review, Decline) and assigns risk tiers (A–E) for easier interpretation. It also generates reason codes to explain the key factors influencing each prediction. A Streamlit application provides an interactive interface for real-time scoring of applicants. Overall, the project demonstrates a full machine learning lifecycle from data preprocessing to deployment, with a focus on interpretability and business applicability.