By Thomas McKeeff & John Kowalski
By extracting meaningful usage patterns, embedded in six months of data, this project helps illustrate and predict service opportunities for Hubway, allowing for an increase in ridership and revenue, while guaranteeing a more enjoyable bike experience for customers.
One appealing feature of Hubway is that users frequently engage in “one-way rentals”, picking up a bike from one station and returning it to a different station. While incredibly convenient for the rider, this poses a unique and continuous problem for Hubway. The pickup and dropoff usage of these bikes is not distributed evenly across locations, which results in some stations often remaining empty without any bikes to rent, while others are full/packed, unable to accept any bike returns. To alleviate this problem, Hubway utilizes rebalancing vans that are dispatched to a full station, in order to move some of the bikes to an empty or nearly empty station. This process helps to decrease the total duration of full or empty stations, but a problem with rebalancing is that the dispersion of the bikes often appears completely random. Because of this unpredictability, the rebalancing vans usually respond only after stations are full or empty, increasing the total duration that these stations are unusable.
Here, station capacity is analyzed as a function of time of day and stations that require attention, whether 'packed stations' (red circles) or 'empty stations' (black circles), are indicated on the map. Circle diameter is based on the number of days that stations were unusable, within a 15 minute time bin, across the 180 days analyzed beginning in April 2012. Over the course of a day, the visualization makes it apparent which stations require attention to minimize service outages and to help maximize the efficiency and effectiveness of the rebalancing vans. Additional analyses (not in the display) show how the usage patterns change dramatically when past weather records indicate it rained on a particular day or when a major event was held in the city (e.g. Red Sox game). These findings can be incredibly helpful in allowing Hubway to better predict complex usage patterns and to dispatch repositioning vans before station service outages occur.
The interactive map is available at http://www.wjh.harvard.edu/~mckeeff/hubway/
Additional download: hubway.pdf