Loan Prediction API with 88.62% accuracy

Loan Prediction API

The Loan Prediction API is a production-ready artificial intelligence system designed to facilitate loan approval predictions through a streamlined, lightweight REST API. It incorporates multiple trained machine learning models, enhancing accuracy to 88.62%. The API is accessible for public use via a demo interface and features comprehensive documentation.

Key Features

• High Accuracy: Supports multiple models to ensure reliable predictions.
• RESTful API Design: Simple integration capabilities for developers.
• Interactive Documentation: Explore the API with Swagger UI for ease of use.
• Input Validation: Validates application data to improve prediction quality.
• Test Coverage: Maintains an 83% coverage to ensure code reliability.

API Overview

• Base URL:
  • Development: http://localhost:5000
  • Production: Live Demo
• Endpoints:
  • GET /: Retrieve API information and statistics.
  • POST /predict: Submit loan application details for prediction.
  • GET /history: Access recent prediction history.
  • GET /statistics: Fetch overall API statistics.

Example Usage

Make a Prediction

Submit loan details to receive a prediction with confidence levels:

{
  "ApplicantIncome": 5000,
  "CoapplicantIncome": 1500,
  "LoanAmount": 150,
  "Loan_Amount_Term": 360,
  "Credit_History": 1,
  "Gender": "Male",
  "Married": "Yes",
  "Dependents": "0",
  "Education": "Graduate",
  "Self_Employed": "No",
  "Property_Area": "Urban"
}

Response Example

The prediction will return a success flag, the predicted status of the loan, confidence, and probability distribution:

{
  "success": true,
  "prediction": "Approved",
  "confidence": 0.89,
  "probability": {
    "rejected": 0.11,
    "approved": 0.89
  }
}

Technology Stack

• Backend: Flask 3.x with Gunicorn
• Database: PostgreSQL for production, SQLite for development
• Machine Learning: scikit-learn, pandas, numpy
• Testing Framework: pytest

Deployment and Development

Effortlessly deploy the API using services like Render, enabling continuous integration directly from GitHub. Developers can run the API locally using Python and Docker, facilitating a seamless development process.

Additional Resources

• API Documentation: Explore here
• Live Demo: Test operational capabilities interactively via this link
• Technical Blog: A detailed technical breakdown of the project journey can be found in a comprehensive medium post.

The Loan Prediction API aims to improve financial decision-making processes through data-driven insights, making the loan approval process more efficient and accessible.