EnergyIn progress
PJM Energy Grid Analytics
A Streamlit app over the PJM grid, the largest wholesale electricity market in the US, with a SHAP-explained classifier that flags conditions likely to produce price spikes.
SHAP
explainable price-spike classifier on live grid data
The question
Wholesale electricity prices on the PJM grid spike hard and fast. Which combinations of load, weather, and market conditions predict a spike, and can the model explain itself?
Data & pipeline
- Data via the gridstatus Python library, the EIA API, and Open-Meteo weather data
- Streamlit app as the front end for exploration
- Price-spike classifier with SHAP values so every prediction is explainable
- Pivoted off PJM Data Miner 2 after hitting API access restrictions, then rebuilt the ingestion on open sources
Stack
PythongridstatusEIA APIOpen-MeteoStreamlitSHAPscikit-learn