This is pretty interesting. I think it may be slightly overconfident in the sense that the data is still not super granular and there aren't too many other data points of comparison (geographic restaurant data, work data, etc.). However, so much data reports viruses being more likely to spread indoors and in close quarters so the subway makes perfect sense.
It is also very scary. I currently live in a fairly remote town of 3000 people. There have been a couple Covid cases here, but it is pretty easy to see how even with the most loose social distancing policies the R0 here would be lower than a fairly stringent policy in NYC. I used to live in Manhattan and the number of people I would be a few feet from just walking out of my apartment, walking one block to a grocery store, and coming back up is more people than I am within a few feet of here every week even when restaurants are open. I'm not sure how big cities solve this.
Milan: https://www.introducingmilan.com/metro
Seoul: http://www.seoulmetro.co.kr/en/cyberStation.do
And the flip side:
Los Angeles: http://www.orangesmile.com/common/img_metro_maps/los-angeles...
San Francisco: http://mapsof.net/san-francisco/san-francisco-bay-area-metro...