Insurance Premium Predictor – Backend (FastAPI)

This is the backend service for the Insurance Premium Predictor project.
It is built using FastAPI and serves a trained Random Forest Classifier model that predicts whether a person’s insurance premium will be Low, Medium, or High.

The backend exposes a /predict API endpoint, which receives user input and returns the prediction in JSON format.

---

🚀 Features

• FastAPI for building a lightweight and fast backend
• Random Forest Classifier trained on user details
• Pydantic for request validation
• Pickle model loading (model.pkl)
• Deployed on Railway

---

🌐 Live Backend

The backend is hosted on Railway and connected to the frontend.
👉 Backend API Base URL

Main endpoint: POST /predict

---

📌 API Usage

Endpoint

POST https://web-production-2c19.up.railway.app/predict

Request Body (JSON)

{ "age": 31, "weight": 91, "height": 1.72, "income_lpa": 10, "smoker": false, "city": "Jaipur", "occupation": "business_owner" }

Response (JSON)

{ "response": { "predicted_category": "Low", "confidence": 0.51, "class_probabilities": { "High": 0.13, "Low": 0.51, "Medium": 0.36 } } }

---

🛠️ Tech Stack

• FastAPI: Web Framework
• Scikit-learn: Random Forest Classifier (trained in Google Colab, saved as model.pkl)
• Pydantic: Request and Data (Input + Output) Validation
• Uvicorn: ASGI server
• Railway: Deployment

---

📌 Related Repositories

• Frontend Repo (Streamlit UI) – Insurance Premium Predictor – Frontend
• Complete Deployed App (Frontend + Backend connected) – FastAPI + ML model code:
  👉 Live App