GridGuard

I was inspired by the 2021 Texas energy crisis, where millions of people went without power in the freezing winter, GridGuard was created to help communities plan around potential energy disruptions during unforeseen events.

GridGuard predicts potential power outages by county using a trained scikit-learn model. Users can easily see which areas have stable, unstable, or critical grids, helping them plan and prepare before an upcoming event.

Features

Interactive map of counties color-coded by grid stability.

State rankings based on predicted affected customers.

AI-generated explanations for why a grid may become unstable.

Built in Streamlit, with predictions powered by a trained linear regression/ensemble model.

Data sourced from public datasets to align with time, location, event type, and voltage logs.

Learning Outcomes

Built a simple predictive model from scratch.

Next Steps

Finding more datasets to train the model and increase prediction accuracy.

Built With

• csv
• gemini
• numpy
• pandas
• plotly
• pydeck
• python
• scikit-learn
• sql
• streamlit