FastAPI ML Application

A machine learning API built with FastAPI that predicts insurance premium categories based on user data using RandomForestClassifier.

Features

• FastAPI backend with ML model integration
• Streamlit frontend for easy interaction
• Dockerized application with multi-service setup
• Health check endpoints for monitoring
• Detailed prediction responses with confidence scores

Tech Stack

• Backend: FastAPI, Python 3.11
• Frontend: Streamlit
• Containerization: Docker
• Process Management: Supervisor
• ML Model: RandomForestClassifier (scikit-learn)

Quick Start

Running with Docker

# Pull the Docker image
docker pull adityaxxz/fastapi-ml-api:latest

# Run the container
docker run -p 8000:8000 -p 8501:8501 adityaxxz/fastapi-ml-api:latest

Running Locally

1. Create and activate a virtual environment:

I've used uv coz its rust-based and 100x faster than pip

uv venv
.venv/bin/activate

2. Install dependencies:

uv pip install -r requirements.txt

3. Start the FastAPI server:

uvicorn app:app --reload
or 
py -m uvicorn app:app --port=8000--reload

4. In a separate terminal, start the Streamlit frontend:

streamlit run frontend.py

API Endpoints

• GET /: Home endpoint

• GET /health: Health check endpoint

• POST /predict: Prediction endpoint

• Swagger UI: http://localhost:8000/docs

Prediction Input Schema

{
  "age": 30,
  "weight": 65.0,
  "height": 1.7,
  "income_lpa": 10.0,
  "smoker": false,
  "city": "Mumbai",
  "occupation": "private_job"
}

Prediction Response Schema

{
  "predicted_category": "Medium",
  "confidence": 0.85,
  "class_probabilities": {
    "Low": 0.10,
    "Medium": 0.85,
    "High": 0.05
  }
}

Project Structure

fastapi-ml/
├── app.py                 # FastAPI application
├── frontend.py            # Streamlit frontend
├── model/
│   ├── predict.py         # Prediction logic
│   └── model.pkl          # Serialized ML model
├── schema/
│   ├── user_input.py      # Input validation schema
│   └── prediction_response.py  # Response schema
├── config/
│   └── city_tier.py       # City tier configuration
├── dockerfile             # Docker configuration
└── requirements.txt       # Python dependencies
``