HELOC Credit Risk Screening Tool

Overview

This project develops a Machine Learning-based Decision Support System (DSS) for preliminary screening of Home Equity Line of Credit (HELOC) applications.

The system predicts whether an applicant is likely to be low-risk (Good) or high-risk (Bad) and provides interpretable explanations for decisions.

Model Information

• Model Type: Logistic Regression
• Accuracy: 71.85%
• ROC-AUC: 0.79
• Features Used:
  • ExternalRiskEstimate
  • NumInqLast6M
  • NetFractionRevolvingBurden
  • NumSatisfactoryTrades
  • AverageMInFile

Deployment

The application is deployed using Streamlit Community Cloud.

How It Works

1. User enters applicant credit information.
2. The model predicts approval probability.
3. If risk is high, the system provides:
   • Main risk reasons
   • Suggested improvement steps

Files Included

• app.py → Streamlit web application
• heloc_model.pkl → Trained Logistic Regression model
• requirements.txt → Required Python packages

Disclaimer

This tool is for educational and preliminary screening purposes only. Final lending decisions should involve professional credit evaluation.