Hubway Data Visualization Challenge

Zach Needell

About:

This is a bunch of loosely-related plots and graphics capturing some of the interesting qualities of Hubway ridership in 2016. The general goal is to capture some of the spatial and temporal patterns of bike sharing in Boston to get a better idea of how Hubway fits in with the Greater Boston transportation network.

First, to get a sense of what the demand looks like, below is a plot of each station's ridership over an average day. The width of each circle represents the total number of trips starting or ending there, and the color represents the ratio of arrivals to departures (arrivals are red, departures are blue). Clicking on any station pulls up a more detailed picture of that station's activity patterns in the box below the map. The two lines represent the average number of arrivals and departures per hour over the given time period.

Change the time period being shown with the dropdowns below the map. Season controls the season of 2016, day of the week allows for toggling betwen weekdays and weekends.

Time of Day:

Season:    
Day of the Week:    

Selected Station:

Station Similarity:

One (perhaps) interesting question about station usage patterns concerns their similarity--which stations show the most similar and divergent travel patterns throughout the day, and how are they spread across the Boston area. Below is a representation of the correlation matrix between the stations' activity patterns. Each square represents the similarity between a pair of stations--purple pairs show similar patterns throughout the day, orange pairs show divergent temporal patterns. Clicking on a pair brings their information to the plot box above, each represented by a different color. Now, the area above 0 represents the average number of arrivals per hour, and below zero represents departures. Stations are ordered to maximize similarity between adjacent stations.

Similarity in temporal patterns is not the only way of linking similar stations. Change the dropdown box below to Connectivity, and station pairs are now colored based on how often trips are taken between them. Stations are ordered based on the Fieldler vector--a method to arrange them in a way to come as close as possible to breaking the stations into disjoint sets (the Eastie stations come close!).

Distance Matrix:    

Monte Carlo

One other question that I kept coming back to was--how long would you expect it take for a bicycle to visit every station? This can be thought of as a Markov process, which is made up of a bunch of states (here: Stations) and transitions between them (here: Trips). The likelihood of each transition depends only on the current state--or, here, where each bike goes next only depends on where (and when) it is now, not where it was before. Below is a simple Monte Carlo simulation of a single bike traveling across the Hubway network. Each trip, the bike randomly chooses the next station based on the observed trip probabilities from 2016. (Note: It gets stuck sometimes, if so hit play again).
Start Simulation:    
Number Visited:

Thanks

CSS is minimal-ui by Susie Lu. Map tiles are from Nextzen. All maps and visualizations are done in d3, with much help from all of Mike Bostock's tutorials. All bugs are my own--feel free to let me know if you find any. Moving map initially inspired by this map of Job Locations.

Contact: zneedell@gmail.com