Loan Approval Prediction 🏦✅

An end‑to‑end machine learning project that predicts whether a loan application should be approved based on applicant details. The model is trained on the LoanApproval.csv dataset using 9 key features:

• Dependents
• Education
• Self_Employed
• ApplicantIncome
• CoapplicantIncome
• LoanAmount
• Loan_Amount_Term
• Credit_History
• Property_Area The project integrates data science, machine learning, and full‑stack deployment with:
• Random Forest Classifier with GridSearchCV for hyperparameter tuning
• FastAPI backend with Uvicorn server
• Frontend UI for user interaction
• Docker for containerization
• CI/CD pipeline for automated testing and deployment
• pytest for unit and integration testing

📌 Project Objective

The goal is to build a loan approval prediction system that:

• Analyzes applicant data to predict loan approval status
• Provides an intuitive web interface for users and loan officers
• Ensures reproducibility and scalability with Docker and CI/CD

🧪 Methodology

• Data Understanding – Explore the LoanApproval.csv dataset
• EDA – Identify trends, correlations, and anomalies
• Feature Engineering – Encode categorical variables, scale numerical features
• Model Training – Random Forest with GridSearchCV for hyperparameter tuning
• Evaluation – Accuracy, Precision, Recall, F1‑score, ROC‑AUC
• Deployment – FastAPI backend + HTML/CSS frontend
• Testing – Unit tests with pytest
• CI/CD – Automated build, test, and deploy pipeline
• Containerization – Docker for consistent environments

📂 Project Structure

├── data/
│   ├── raw/                  # Original dataset(s)
│   ├── processed/            # Cleaned & transformed datasets
│
├── frontend/
│   ├── index.html            # Welcome page
│   ├── predict.html          # Loan prediction form
│   └── result.html           # Display prediction results
│
├── notebooks/
│   ├── 01_eda.ipynb
│   └── 02_model_building.ipynb
│
├── scripts/
│   ├── preprocessing.py       # Clean and transform data
│   ├── train_model.py         # Train and return model
│   ├── predict.py             # Predict Loan Approval
│   ├── evaluate.py            # Evaluate and save results
│
├── models/                    # Saved ML models (.pkl)
│
├── test/
│   └── test_code.py           # pytest unit tests
│
├── results/
│   ├── training_testing_results.csv
│   ├── feature_importance.csv
│   └── best_parameter.csv
│
├── .dockerignore
├── Dockerfile
├── main.py                    # FastAPI app entry point
├── requirements.txt
├── README.md
└── .gitignore

🖥️ Frontend

• index.html → Welcome page with project intro
• predict.html → Form for users to input loan details (9 features)
• result.html → Displays prediction result (Approved/Rejected)

⚙️ Backend

• FastAPI app (main.py) exposes endpoints:
• / → Welcome page
• /predict → Loan approval prediction
• Uvicorn runs the ASGI server

🐳 Docker Support

A Dockerfile is included to containerize the app:

from python:3.9-slim

# Set work directory
WORKDIR /app
# Install system dependencies
RUN apt-get update && apt-get install -y \
    build-essential \
    && rm -rf /var/lib/apt/lists/*

# Copy requirements file
COPY requirements.txt /app/

# Install Python dependencies
RUN pip install --no-cache-dir -r requirements.txt

# Copy project source code
COPY . /app/

# Expose port for FastAPI
EXPOSE 8000

# Command to run FastAPI with Uvicorn
CMD ["uvicorn", "main:app", "--host", "0.0.0.0", "--port", "8000", "--reload"]

Build and run:

docker build -t loan-approval-app .
docker run -p 8000:8000 loan-approval-app

🔄 CI/CD Pipeline

• Implemented with GitHub Actions
• Workflow includes:
• Install dependencies
• Run pytest
• Build Docker image
• Deploy to cloud (AWS, Azure, or GCP) Example workflow file: .github/workflows/ci.yml

name: CI Pipeline

on: [push, pull_request]

jobs:
  build-test:
    runs-on: ubuntu-latest
    steps:
      - uses: actions/checkout@v3
      - name: Set up Python
        uses: actions/setup-python@v4
        with:
          python-version: "3.10"
      - name: Install dependencies
        run: pip install -r requirements.txt
      - name: Run tests
        run: pytest

🧪 Testing

• pytest used for unit and integration tests
• Example test (test/test_code.py):

from scripts.predict import predict_loan

def test_predict_loan():
    # Example input: [Dependents, Education, Self_Employed, ApplicantIncome, CoapplicantIncome, LoanAmount, Loan_Amount_Term, Credit_History, Property_Area]
    result = predict_loan([1, "Graduate", "No", 5000, 2000, 150, 360, 1, "Urban"])
    assert result in [0, 1]   # 0 = Rejected, 1 = Approved

Run tests:

pytest

🚀 End-to-End Flow

• User opens frontend in browser
• Inputs loan application details into form
• Request sent to FastAPI backend
• Backend loads trained Random Forest model
• Model predicts loan approval (Approved/Rejected)
• Result displayed on result.html
• CI/CD ensures continuous testing and deployment
• Docker ensures reproducibility across environments

✅ Key Features

• End‑to‑end ML pipeline (EDA → Model → Deployment)
• Random Forest with GridSearchCV for hyperparameter tuning
• FastAPI backend with REST endpoints
• Frontend UI for user interaction
• Dockerized for portability
• CI/CD pipeline for automation
• Unit tests with pytest