Credit risk model for loan applicants

The brief

A non-bank lender approves loans with a scorecard nobody has checked in years. Build a model that predicts the chance of default from the application, compare it with the simple scorecard, and explain each decision so a credit officer can review it. Good looks like: careful handling of missing data, honest validation, and explanations for approved and declined cases.

What you will build

1. Explore and clean the application data
2. Engineer features such as debt-to-income ratios
3. Train logistic regression and decision trees; compare with a simple scorecard
4. Evaluate with AUC and the cost of bad loans
5. Explain individual decisions

Free, open data to build it with

• Home Credit Default Risk ↗ Kaggle
• Statlog (German Credit Data) ↗ UCI Machine Learning Repository

Check each dataset's licence before you publish your work. Your mentor gives your team its own version of the brief (data, scope or region), so every team's project is its own.