conquest-rgb/grid-sense

Power outage forecasting for Kenya # grid-sense Power outage forecasting for Kenya # GridSense: Short-Horizon Power Outage Forecasting Early warning system for power outages in Kenya, helping businesses and utilities prepare for disruptions 6-24 hours in advance. --- ## Project Overview **GridSense** predicts power outages and identifies maintenance risks using: - Weather data (rainfall, wind, storms) - Grid topology from OpenStreetMap - Satellite night-lights for validation - Scraped planned outage notices ### Why This Matters - **Businesses**: Reduce diesel costs, prevent spoilage, optimize operations - **Critical Infrastructure**: Hospitals, data centers, telecom towers can prepare - **Utilities**: Target maintenance and vegetation clearing --- ## Features - **6-24 Hour Outage Predictions** at county/substation level - **Calibrated Probabilities** (not just yes/no) - **Planned Outage Dashboard** with clean locations and times - **Maintenance Risk Signals** for utility operators - **Interactive Map** showing risk levels by region --- ## Project Structure ``` grid-sense/ ├── config/ # Configuration files ├── data/ # Data directory (not tracked in git) │ ├── raw/ # Scraped & downloaded data │ ├── processed/ # Cleaned & engineered features │ ├── external/ # Third-party datasets │ └── labels/ # Validation labels ├── dashboard/ # Streamlit web application │ ├── app.py # Main dashboard │ └── pages/ # Multi-page components ├── docs/ # Documentation ├── models/ # Trained models (not tracked) ├── notebooks/ # Jupyter notebooks for analysis ├── outputs/ # Generated figures and reports ├── scripts/ # Utility scripts │ ├── scrape_outages.py │ ├── download_weather.py │ └── train_model.py ├── src/ # Core source code │ ├── data/ # Data processing modules │ ├── models/ # ML models │ ├── features/ …