suvoganguli/urban-risk-pipeline

ML project showcasing usage of open-source data to identify urban vulnerability areas focusing on Kampala, Uganda Urban Vulnerability Pipeline ============================= This project identifies and visualizes urban vulnerabilities by integrating geospatial and socioeconomic data. The pipeline combines: - **OpenStreetMap (OSM)** infrastructure layers (buildings, roads, services) - **Sentinel-2 NDVI** satellite imagery - **WorldPop** demographic estimates The final output includes composite vulnerability scores across a 2×2 km grid, supporting applications in disaster preparedness, equitable development, and urban planning. > ⚠️ Note: While the repo code uses the term “Urban Risk Pipeline,” the official project name is **Urban Vulnerability Pipeline**. ## 📁 Repository Structure This pipeline is organized into modular notebooks and data folders: ### 📓 Notebooks | Notebook | Description | |----------|-------------| | `01_load_and_explore_osm_data.ipynb` | Loads and explores raw OSM layers (buildings, roads, services) | | `02_process_ndvi_satellite_data.ipynb` | Loads Sentinel-2 NDVI raster, creates a 2×2 km grid, computes zonal NDVI stats | | `03_add_population_from_worldpop.ipynb` | Adds WorldPop population estimates to each grid cell | | `04_kawempe_grid_features_ml.ipynb` | Combines all features, scales them, computes vulnerability scores, and visualizes the results | --- ### 📂 Data Folders | Folder | Contents | |--------|----------| | `data/raw/` | Original input files (OSM GeoJSONs, NDVI raster, WorldPop raster) | | `data/processed/` | Intermediate outputs (e.g., `urban_grid.geojson`, `urban_grid_with_population.geojson`) | | `visualizations/` | Saved map visualizations for NDVI, infrastructure, and vulnerability classifications | --- ## 📊 Outputs - **NDVI Distribution Maps** - **Infrastructure Heatmaps** - **Hidden Vulnerability Score** — highlights high-pop, low-infra areas - **Rescue Priority Score** — identifies high-density, well-built areas for disaster response --- ## 🚀 Future Directions - Add time-series vulnerability tracking - Integrate flood risk …