codereport / city-strides-hacking Goto Github PK

That discussion had more turns and bends than the mean output of your current algorithm. I went from "Yes Bryce, tell him to get his graph theory up to snuff." The distances, are the weights. He needs to prune the nodes that pose no choice, keep just the intersections (which naively are the vertices with exactly 2 edges, but some ways change name halfway, so some checks have to be done for Toronto, as you don't want to prune any such public way).
But than Bryce failed to see the difference between the solved shortest path from A-B problem (Dijkstra or A* if you have some useful heuristic) and your NP-hard problem.
And when you mentioned your first implementation was in Python I just had to point you to this Python sketch: https://www.youtube.com/watch?v=VI4q4ShXNYw

• change output from nodes.csv to hot_spots.csv
• display on plot

Goals

1. Figure out how to query OSM
2. Build query for all streets/nodes/edges for local frame/global
3. Build route builder for:
   • Most streets in a new city
   • Most efficient route in a dense area (no nodes left behind)
4. Determine best cities in the world (shortest average street length)

Resources:

• https://download.geofabrik.de/
• https://taginfo.openstreetmap.org/keys
• http://overpass-turbo.eu/

Right now all we have is a 2n * 2n grid and we go to the one with the most nodes. It should really be more granular via maybe a clustering algorithm? and each a target cluster should be weighted by distance and only be completed when the roads are done.

Can be done after #5