Forecast Hourly Energy Demand for a Microgrid Operator
Overview
What this challenge is about.
You receive 24 months of hourly demand, weather (temperature, irradiance), and calendar data for the community. Build a probabilistic forecaster (e.g., quantile regression with gradient boosting, or a deep model like DeepAR) that outputs 48h-ahead p10/p50/p90 quantiles. Evaluate on rolling-origin holdouts with pinball loss + coverage. Build a Streamlit dispatcher view showing the next 48 hours with the interval shaded. Deliver model, evaluation, dispatcher UI prototype, and a 3-page integration memo.
The Brief
What you'll do, and what you'll demonstrate.
Build a 48h probabilistic energy-demand forecaster with calibrated intervals and a dispatcher-friendly UI.
Earning criteria — what you'll demonstrate
- Build probabilistic forecasts (not just point estimates)
- Evaluate forecasts with pinball loss and coverage
- Translate interval forecasts into a usable dispatcher UI
- Document the integration with grid-import bidding
Program Fit
Where this fits in your program.
Sharpens the same skills your degree expects you to demonstrate.
Skills
Skills you'll demonstrate.
Each one shows up on your verified credential.
Careers
Roles this prepares you for.
Real titles. Real skill bridges. Pick the one closest to your trajectory.
Machine Learning Engineer
Probabilistic forecasting with calibrated intervals and a shipped operator UI is exactly the day-to-day work of MLEs in energy and climate tech.
This challenge sharpens
- probabilistic-forecasting
- quantile-regression
- calibration
Data Scientist
Pinball-loss-based evaluation with sliced coverage is bread-and-butter senior data-science work on forecasting problems.
This challenge sharpens
- probabilistic-forecasting
- calibration
- quantile-regression
AI Engineer
Wiring a probabilistic forecaster into a dispatcher tool is the integration craft that AI engineers ship at climate-tech startups.
This challenge sharpens
- dashboard-design
- deep-forecasting
- python