Optimal MBTA Green Line Stops

We were motivated by a belief that the MBTA’s Green Line stops too frequently. This interactive map shows optimal stops based on different calculated k-means scores for clustering the Green Line T stops in Boston. The variant of the clustering algorithm takes into account a given weight for each point, and the weight is a function of two primary characteristics: the stop’s current popularity and the proximity to the nearest alternative stop (measured by walking distance).

Initally, the map shows the current existing Green Line stops (where k = 0). Call those x. By playing with the sliders and changing the value of k, you can see the optimal k stops for that branch, suggesting all other x - k are the ones to eliminate if aiming for that level of scaling back.

     GLB      GLC      GLD      GLE

k Sliders

Each of the original stops are represented by a transparent circle of corresponding color (see key up top). Each optimal stop appears as a solid circle of the same color with a black border.

In 2014, the MBTA proposed removing four stops from the B line: BU West, St Paul St., Pleasant St., and Babcock St., replacing them with one new stop just west of BU West and a second between Babcock St. and Pleasant St. The results here agreed; if we were to use this experiment to improve the Green Line B branch by just one stop, it would remove the Pleasant St. and Babcock St. stops, replacing them with a new stop in between. So that’s pretty cool.

For details and all source code, see the project on GitHub. For similar projects, see the Data Mechanics site.