Electricity demand forecast

The brief

Load dispatch centres buy power a day ahead, and forecasting errors cost money. Forecast daily (or hourly) electricity demand with weather and calendar effects, and prove it beats a seasonal baseline. Good looks like: careful features, back-testing over a year, and error measures a dispatch engineer uses.

What you will build

1. Prepare demand series and calendar features
2. Add weather and holiday features
3. Compare seasonal-naive, Prophet and gradient-boosted forecasts
4. Back-test over a full year
5. Report errors by season and on holidays

Free, open data to build it with

• Hourly energy consumption (PJM) ↗ Kaggle
• Grid-India reports ↗ Grid Controller of India

Check each dataset's licence before you publish your work. Your mentor gives your team its own version of the brief (data, scope or region), so every team's project is its own.