Post by Christopher Fox

🚶‍♂️I built a walkability index for Toronto using a GIS ETL pipeline - and now I’m wondering how accurate it actually feels on the ground. This project was mainly about practicing GIS ETL using R, turning raw geospatial data into something clean and reusable. 🔍 Extract: • Neighbourhood boundaries (GeoJSON) • 64,000+ sidewalk geometries • 196,000+ OpenStreetMap amenities (food/drink, daily services, schools & libraries, shops, parks/fitness, transit) 🔄 Transform: • Fixed CRS + geometry issues so distance calculations worked • Cleaned and standardized messy fields • Counted amenities and sidewalk density per neighbourhood • Built a normalized (0–1) walkability index 💾 Load: • Exported a clean GeoPackage + CSV for future analysis. The result shows Toronto’s downtown core as highly walkable, with scores dropping farther out. If I iterate on this, I’d love to push it toward intersection density, weighted amenities, and network-based walking distances instead of polygon level aggregation. Neighbourhood averages smooth out a lot of real world variation. If you live in Toronto, does this match your experience walking around the city? Any areas that feel off? (Also thinking about trying this approach on Halifax next! 👀) Choropleth Map produced in R using sf + ggplot2 #GIS #SpatialData #UrbanPlanning #RStats #Toronto