The winners have been announced, and the challenge is over. The winning entry did something very clever. They used MBTA data and calculated the amount of time saved by using Hubway versus public transit. The author recognizes that it’s hard to get meaningful conclusions about transit when you look at things in isolation. If I had infinite time, here are some other analyses I’d like to run.
Predicting Optimal Hubway Locations
Really what Hubway wants to know is where to put their stations. It’s a hard thing to pin down. One could look at population density easily enough (as I attempted to do), but that leaves out the commercial aspects. There may be data available on where people work, something like jobs/city block as opposed to occupants, that would be really outstanding. Alternatively one could estimate based on tax receipts, zoning, etc. Do-able, but a big task.
Hubway Network Analysis
Hubway has to rebalance the stations, because the trips people take aren’t symmetric. Preventing this from being done would save a lot of effort. The easy thing to do is look at temporary imbalances. For example, lots of people seem to go from North Station to South Station in the morning, and back in the afternoon. If each station is given enough excess capacity it means a lot less rebalancing .
Another option would be to try to place stations so that the network is more balanced. Maybe people commute home -> work in the morning, but maybe run errands after work. So if one could figure out where to place stations to essentially function as bridges, or excess capacity, for others, that could help out the network overall. The more I think about this the more it just seems like putting new stations where they’d get used. Maybe it is, but it could be a new approach to coming up with possible locations.
Overall Transit Network Analysis
Walk. Bike. Bus. Train. To get from point A to point B you can use any of these options. Hubway one place, get a bit tired, take a bus, and so forth. The MBTA has ridership data; I’m not sure if walking data is available (it certainly wouldn’t be exact, but maybe survey data exists). I don’t think you could track individuals (since Hubway is a separate system from MBTA, paying for Hubway with a Charliecard would be convenient and solve this problem btw), but one could track overall flow of people. Locations which use the MBTA heavily, but don’t have enough Hubway capacity, would be prime targets for new Hubway bikes. Or vice versa for additional bus routes or more frequent trips.
-  Of course, that excess capacity has its own cost ↩