MBTA Capstone Project Creating a better revenue model

Transfer Network

Below is a network of all of the transfer information we recovered from the Charliecard data. Each node represents a route and each directed edge represents the volume of transfers from one route to another. Nodes are sized based on their weighted in-degrees, and colored by a rudimentary community detection algorithm using the modularity score.

The network highlights bus routes that feed the transit system. Many customers have to take a bus to get to the subway system. For example, the 111 is one of the highest transfer-volume busses, bringing in large amounts of work commuters from the Chelsea area. The average number of segments per journey is 1.3.

The clustering seems to suggest strong relationships between routes - one example is Ashmont and Jackson Square, which are not particularly geographically close, but are two hubs in two distinct regions that aren't well connected by train.

One limitation of this data, and thus this visualization, is that our swipe data only includes subway stations that customers got on the train, but not off the train, which means that the transfer we see on the return is different. For example, if I take the Red line from Davis Square to Harvard Square, and switch to the 66, then it would show up as a transfer from Davis Square to the 66. On the return, it would show up as a transfer from the 66 to Harvard Square.