Project lead: David Marulli (@dmarulli | [email protected])
Primary developer: @YukunVVan
Orgs: Streets Data Collaborative | ARGO
Imagine looking at any non-cartographical data visualization or UI component (tables, graphs, text, etc.) that aims to display, say, characteristics of the worst potholes in the city. To represent the location, we could display a coordinate (e.g. 40.7217267,-73.9870392), but without looking at a map, where is this?
One could imagine instead displaying an address (i.e. 263 E Houston St), but, again, without looking at a map, even to a native New Yorker like myself, I don't know immediately know if this is on the east or west side of the city.
Building on a solid foundation, the deliverable of this project will programmatically output more immediately meaningful descriptions (i.e. "Houston Street between Avenue B and Avenue C").
As folks who are in the data weeds day-to-day, it's important to not lose sight of higher-level user experience.
OSMnx is a python package that allows one to pull street networks for cities around the world. geopandas is a python package for processing geospatial data. Using these two packages as our foundation, we should be able to create a tool that recieves a lat/lng pair and returns a conversational string, such as "Houston Street between Avenue B and Avenue C". Below are the major steps we follow:
- Use OSMnx to get a street graph
- Given a lat/lng point, query for nearest street segment
- Get to/from node IDs of street segment, and query for intersecting streets
- Compare street names of intersecting streets with original segment to determine to/from streets
