Credit Risk & Loan Default Prediction

A complete end-to-end machine learning project for predicting credit risk and loan defaults. This project demonstrates data preprocessing, feature engineering, model training, evaluation, and deployment via a REST API.

Table of Contents

• Overview
• Features
• Project Structure
• Installation
• Quick Start
• Usage
  • Training
  • API
• API Endpoints
• Model Performance
• Configuration
• Testing
• Contributing
• License

Overview

This project implements a complete machine learning pipeline for credit risk assessment:

• Data Preprocessing: Handle missing values, scale features using StandardScaler
• Feature Engineering: Create derived features like debt-to-income ratio, credit utilization
• Model Training: Train and compare multiple ML models (Logistic Regression, Decision Tree, Random Forest, Gradient Boosting)
• Model Evaluation: Evaluate with various metrics (Accuracy, Precision, Recall, F1, ROC-AUC)
• Deployment: REST API using Flask for real-time predictions

Features

• Modular architecture with separate components
• Configuration-based settings (YAML)
• Comprehensive logging
• Unit tests
• Makefile for common commands
• Professional package structure

Project Structure

credit-risk-loan-default-prediction/
├── app/
│   └── app.py                  # Flask REST API
├── config.yaml                 # Configuration file
├── data/
│   ├── processed/              # Processed data
│   └── raw/                    # Raw dataset
├── Makefile                    # Make commands
├── models/                     # Trained models
├── notebooks/
│   └── eda.ipynb               # Exploratory Data Analysis
├── README.md
├── requirements.txt
├── setup.py                    # Package setup
├── src/
│   ├── __init__.py             # Package initialization
│   ├── config.py               # Configuration loader
│   ├── data_generator.py       # Data generation
│   ├── data_preprocessing.py   # Data preprocessing
│   ├── feature_engineering.py # Feature engineering
│   ├── logging_config.py       # Logging setup
│   ├── model_evaluation.py     # Model evaluation
│   ├── model_training.py       # Model training
│   ├── train.py                # Main training script
│   └── utils.py                # Utility functions
└── tests/                      # Unit tests
    ├── __init__.py
    └── test_data_preprocessing.py

Installation

Option 1: Install from source

# Clone the repository
git clone https://github.com/logeshkannan19/Credit-Risk-Loan-Default-Prediction.git
cd Credit-Risk-Loan-Default-Prediction

# Create virtual environment
python3 -m venv venv
source venv/bin/activate  # On Windows: venv\Scripts\activate

# Install dependencies
pip install -r requirements.txt

# Install package in development mode
pip install -e .

Option 2: Using Make

make install

Quick Start

1. Train the Model

python -m src.train

Or using Make:

make train

2. Start the API

python -m src.app

Or:

make api

The API will start on http://localhost:5000

3. Make a Prediction

curl -X POST http://localhost:5000/predict \
  -H "Content-Type: application/json" \
  -d '{
    "age": 35,
    "income": 75000,
    "credit_score": 650,
    "debt_amount": 15000,
    "monthly_expenses": 2500,
    "employment_years": 5,
    "loan_amount": 25000,
    "existing_credits": 2,
    "interest_rate": 12.5,
    "payment_history": 85
  }'

Usage

Training

The training script can be run with:

python -m src.train

This will:

1. Load and preprocess the data
2. Engineer new features
3. Train multiple models
4. Evaluate and select the best model
5. Save the model and preprocessing artifacts

API

Start the Flask API:

python -m src.app

API Endpoints

Endpoint	Method	Description
/health	GET	Health check
/predict	POST	Single prediction
/batch_predict	POST	Batch predictions
/model_info	GET	Model information

Example Response

{
  "prediction": "No Default",
  "probability": 0.4235,
  "risk_level": "Medium"
}

Risk Levels

• Low: Probability < 0.3
• Medium: 0.3 <= Probability < 0.6
• High: Probability >= 0.6

Model Performance

The model is evaluated using multiple metrics:

Model	Accuracy	Precision	Recall	F1 Score	ROC-AUC
Logistic Regression	0.4947	0.6089	0.5014	0.5500	0.4989
Decision Tree	0.5214	0.6126	0.6061	0.6094	0.4946
Random Forest	0.5506	0.6166	0.7142	0.6618	0.5085
Gradient Boosting	0.5973	0.6139	0.9320	0.7403	0.5068

Best Model: Random Forest (selected based on ROC-AUC score)

Configuration

All settings can be configured in config.yaml:

# Training Settings
training:
  test_size: 0.2
  random_state: 42
  cv_folds: 5

# API Settings
api:
  host: "0.0.0.0"
  port: 5000

# Risk Thresholds
risk:
  low: 0.3
  medium: 0.6

Testing

Run tests with:

pytest tests/ -v

Or using Make:

make test

Contributing

1. Fork the repository
2. Create a feature branch (git checkout -b feature/amazing-feature)
3. Commit your changes (git commit -m 'Add amazing feature')
4. Push to the branch (git push origin feature/amazing-feature)
5. Open a Pull Request

License

This project is licensed under the MIT License - see the LICENSE file for details.

Author

• Logesh Kannan - logeshkannan19

Acknowledgments

• Scikit-learn for machine learning utilities
• Flask for REST API
• The "Give Me Some Credit" dataset inspiration