Skip to content

README.md

Latest commit

 

History

History
214 lines (144 loc) · 3.42 KB

README.md

File metadata and controls

214 lines (144 loc) · 3.42 KB

🚚 ETA Prediction System (End-to-End ML Engineering Project)

Production-ready machine learning system for real-time delivery ETA prediction with API, deployment, UI, authentication, and analytics.

🌐 Live Demo

📌 Overview

This project simulates a real-world food delivery ETA prediction system similar to Swiggy/Zomato.

It goes beyond model building by implementing a complete ML system:

  • Model training
  • API serving
  • Docker containerization
  • Cloud deployment
  • Interactive frontend
  • Logging & analytics

🧠 Problem Statement

Predict the Estimated Time of Arrival (ETA) for a delivery based on:

  • Distance
  • Order time
  • Traffic conditions
  • Preparation time
  • Weather
  • Delivery partner rating

🏗️ System Architecture

User → Streamlit UI → FastAPI → ML Model → Prediction
                         ↓
                     Logging → Analytics Dashboard

⚙️ Tech Stack

🔹 Machine Learning

  • Python, Pandas, Scikit-learn

🔹 Backend

  • FastAPI (REST API)
  • Uvicorn

🔹 Frontend

  • Streamlit (interactive UI)

🔹 DevOps

  • Docker (containerization)
  • Render (deployment)

🔹 Data & Analytics

  • JSON logging
  • Pandas analytics dashboard

🚀 Features

  • ✅ Real-time ETA prediction API
  • ✅ Interactive UI (Streamlit)
  • ✅ User authentication (session-based)
  • ✅ Input validation & structured requests
  • ✅ Dockerized application
  • ✅ Cloud deployment (public access)
  • ✅ Logging of predictions
  • ✅ Analytics dashboard (trends & insights)

📊 Sample Input

{
  "distance_km": 3.5,
  "order_hour": 13,
  "traffic_level": 4,
  "prep_time": 20,
  "weather": 1,
  "delivery_rating": 4.5
}

📈 Sample Output

{
  "predicted_eta": 29.7
}

📁 Project Structure

ml-eta-project/
│
├── app/                # FastAPI app
├── src/                # ML training & prediction logic
├── model/              # Saved model
├── logs/               # Prediction logs
├── ui_app.py           # Streamlit UI
├── Dockerfile
├── requirements.txt
└── README.md

🧪 How to Run Locally

1️⃣ Clone repo

git clone https://github.com/RajeshKumar3451/ETA-Engine.git
cd your-repo

2️⃣ Install dependencies

pip install -r requirements.txt

3️⃣ Run API

uvicorn app.main:app --reload

4️⃣ Run UI

streamlit run ui_app.py

🐳 Run with Docker

docker build -t eta-app .
docker run -p 8000:8000 eta-app

📊 Analytics Dashboard

Tracks:

  • Total predictions
  • Average ETA
  • Traffic distribution
  • Prediction trends

🔐 Authentication

Simple session-based login system implemented in Streamlit UI.

💬 Key Learnings

  • End-to-end ML system design
  • API development & deployment
  • Handling semi-structured data (JSON logs)
  • Building user-facing ML applications
  • Monitoring & analytics

📌 Future Improvements

  • Real-time traffic (Google Maps API)
  • Database integration (PostgreSQL)
  • CI/CD pipeline
  • Model versioning
  • Scalable microservices architecture

👤 Author

Rajesh Kumar

⭐ Support

If you found this useful, consider starring ⭐ the repo!