At Faraday, we dig OSM.
OpenStreetmap (OSM) is the foundation of our basemap and a model of the power of open data. It guides customers on our platform to their ideal audiences . . .
. . . and it serves as building blocks for geospatial analysis, both the kind we already do and the kind we want to do more of.
The problem is that it's big. The entire OSM database is portable, but at 50GB it's not very friendly. Sometimes we just want the driveway network of one county, or the building footprints in a zip code. Whole companies have sprung up around this workflow, but we have a few tried-and-true-and-cheap tools that we rely on:
- Mapzen-hosted metro extracts - If your desired zone is on the list of regularly-updated cities, just grab the shapefiles and go!
- OSM vector tiles - Use these with toolsets like tilereduce for distributed geoprocessing at tile scale.
- Overpass API - This tuneable endpoint works great for specific queries in minutely-defined regions (e.g. find all the one-way streets in Park Slope), but it can be a bit opaque. Use the query-overpass node module to spit out GeoJSON with minimal fuss.