AI / Forecasting / Team project · 8 people
The grid’s next hour
24-hour demand forecasting using Denmark DK1 energy data.
The problem
Model calendar, price and weather effects in hourly demand.
The approach
I led algorithm development, using 40+ features, 1–168 hour lags and recursive forecasting.
Results and scope
- R² 0.97; MAE 44.6 MWh on a chronological 80/20 split.