CS224W

How to run code

Download Uber Movement data and put in this folder:
- San Francisco:
  - movement-speeds-quarterly-by-hod-san-francisco-2019-Q2.csv.zip
- New York:
  - movement-speeds-quarterly-by-hod-new-york-2019-Q2.csv.zip
- Seattle:
  - movement-speeds-quarterly-by-hod-seattle-2019-Q2.csv.zip
Install necessary packages with conda env create -f environment.yml
Run download_osm.py to download OSM maps for San Francisco, New York, Seattle.

Possible Next Steps

Get Google Cloud set up.
Get OpenStreetMap nodes/ways as a networkx graph using osmnx (or some other library).

For example: https://github.com/gboeing/osmnx-examples/blob/master/notebooks/06-example-osmnx-networkx.ipynb

Use Uber Movement data to add traffic information to the graph. We have the OpenStreetMap mapping so it should not be hard.
Convert graph from networkx to pytorch_geometric using torch_geometric.utils.from_networkx
Run GCN, GraphSAGE or whatever using pytorch geometric.