Energy Price Forecasting - MLOps Pipeline

Production-ready time series forecasting for European electricity markets with FastAPI, Streamlit, and Google Cloud deployment.

🎯 What This Project Does

• ⚡ Price Forecasting for German, French, and Dutch electricity markets (up to 168 hours)
• 🤖 Multiple ML Models including XGBoost, LightGBM, ARIMA, and ensemble methods
• 🚀 Production API with FastAPI for real-time predictions
• 📈 Interactive Dashboard built with Streamlit
• ☁️ Cloud Deployment on Google Cloud Platform with auto-scaling

🌐 Demo

📺 Demo Video

🚀 How to Use

API Usage

Start the API locally with the steps under Local Development, then:

Get Forecast:

curl -X POST "http://localhost:8000/predict" \
  -H "Content-Type: application/json" \
  -d '{"country": "DE", "model": "xgboost"}'

Health Check:

curl http://localhost:8000/health

Local Development

# Clone and setup
git clone https://github.com/aishwaryaj7/aishwaryaj7.github.io.git
cd aishwaryaj7.github.io/ai-portfolio/projects/energy_price_forecasting

# Install dependencies
pip install -r requirements.txt

# Train models and start API
python train.py
uvicorn api.main:app --reload

# Launch dashboard (separate terminal)
streamlit run streamlit_app.py

🔧 Key Features

• Multi-Market Support: Germany, France, Netherlands
• Flexible Forecasting: 1-168 hour prediction horizons
• Multiple Models: XGBoost, LightGBM, ARIMA, Ensemble
• Real-time API: FastAPI with automatic documentation
• Interactive Dashboard: Streamlit visualization
• Cloud Deployment: Google Cloud Run with Docker
• MLOps Integration: MLflow for experiment tracking

🛠️ Tech Stack

Machine Learning: Scikit-learn, XGBoost, LightGBM, Statsmodels API: FastAPI, Uvicorn, Pydantic Frontend: Streamlit Cloud: Google Cloud Run, Docker MLOps: MLflow CI/CD: GitHub Actions

🤝 Skills Demonstrated

• Time Series Forecasting & MLOps
• FastAPI Development & Cloud Deployment
• Docker Containerization & CI/CD
• Data Engineering & API Design