Back to Projects
PROJECT #02 / 05

Grid Intelligence

Electricity price forecasting for the DE-LU bidding zone

ENERGY · 2026
Live Demo

THE CHALLENGE

Day-ahead electricity prices in the DE-LU market are volatile, driven by renewable generation, cross-border flows, and weather patterns. Energy traders, grid operators, and industrial consumers need accurate forecasts to optimize bidding strategies and manage risk. Traditional models struggle with the non-linear relationships between weather, generation capacity, and price formation.

THE SOLUTION

Grid Intelligence uses machine learning to forecast day-ahead electricity prices for the Germany-Luxembourg bidding zone. The system ingests real-time data from ENTSO-E — historical prices, generation by source, cross-border flows, and planned outages — and engineers features that capture market dynamics: renewable penetration, residual load, and time-of-day patterns.

The model is a gradient-boosted decision tree trained on three years of hourly data, optimized for mean absolute error. A Streamlit dashboard visualizes forecasts alongside actual prices, highlighting accuracy and model confidence. Deployed on GCP with automated daily retraining, the system adapts to shifting market conditions and provides actionable intelligence for energy decision-makers.

TECH STACK

Python XGBoost scikit-learn pandas ENTSO-E API Streamlit GCP Docker