Loan Pricing Optimization

End-to-end regression pipeline for predicting optimal loan amount and interest rate. Built for credit risk and fintech contexts.

Project Overview

Predict loan amount and interest rate for a given borrower profile. Used by underwriting teams to validate pricing decisions and identify outlier quotes.

Business Context

Loan pricing is a core credit function — the right price balances:

• Profitability (rate high enough to cover risk)
• Competitiveness (rate low enough to win business)
• Risk alignment (rate reflects borrower's true default probability)

This pipeline trains regression models to predict:

1. Recommended loan amount (regression)
2. Optimal interest rate (regression)

Files

loan-pricing-optimization/
├── README.md
├── requirements.txt
├── run_pipeline.py          ← entry point
├── src/
│   ├── __init__.py
│   ├── data_loader.py       ← data ingestion + split
│   ├── features.py          ← feature engineering
│   ├── train.py             ← model training + comparison
│   └── predict.py          ← production inference
├── models/                   ← saved artifacts
├── data/                     ← raw data
└── reports/                 ← evaluation reports

Model Performance

Target: R² > 0.70, RMSE < 15% of target mean on test set.

Skills Demonstrated

• Regression (Linear, Ridge, Random Forest, XGBoost)
• Feature importance analysis
• Business metric translation (RMSE, MAE, MAPE)
• Multi-output prediction
• Price sensitivity analysis