A sample of Capital Bikeshare GPS trip data offers a unique look into rider behaviors
This series of posts discusses a study undertaken in the spring of 2015 where several Capital Bikeshare bikes were outfitted with GPS devices and tracked for several weeks.
Most bikeshare programs generate a wealth of data about their trips—where they start and end, what kind of user is taking them. Some, such as Phoenix’s dockless , even generate GPS data for its users to view. Washington, D.C.’s Capital Bikeshare, however, does not, and so virtually nothing is known about where the bikes actually go – what routes they take, where they stop, and what infrastructure they use.
To gather this information, the District Department of Transportation in the spring of 2015 outfitted a number of Capital Bikeshare bikes with GPS devices and tracked them for several weeks. This undertaking resulted in a wealth of data (over 300,000 GPS points), and DDOT approached Virginia Tech’s Urban & Regional Planning program in Alexandria for help in deciphering it all. I took the opportunity to fulfill the capstone project requirements as part of my Master’s Degree in Urban & Regional Planning.
DDOT mounted GPS trackers under the saddles of 130 bikeshare bikes and ultimately recovered 94 trackers. The first bikes were distributed beginning April 20, 2015, and recorded data until they ran out of batteries – the longest lasting until May 16. After significant data cleaning, which removed false trips like rides in a rebalancing van, 3,596 trips worth of GPS data remained and was calculated to be representative of all bikeshare trips within the same time period. Roughly three-quarters of these trips were by member users who hold monthly or annual memberships. The remaining quarter were casual users, those using daily or three-day memberships and tend to be tourists.
Previous studies and visualizations of Capital Bikeshare trips had to assume direct-line routes or suggested bicycle routes in their analyses. Because of this, trip distances are often underestimated, and no additional details can be gleaned from trips that have the same start and end stations, of which there is a large percentage.
The animation below shows actual paths taken. All of the trips were condensed into a single day. Notice the direct routes taken by member users and the casual users touring locations near the National Mall.
[vc_video link=”https://youtu.be/y0sYziuaahU”]
The differences in trip attributes, such as duration, speed, and distance, between member and casual trips were stark. The average casual user trip was nearly twice as long as the member trip in distance, three times longer in duration, and considerably slower in average speed. Comparing actual tracked distance with the direct distance between start and end stations also reveals a huge difference, as member users take much more direct routes. Their trips were 35 percent longer than a straight line, compared with casual trips which were more than 177 percent longer. These figures can be helpful for future bikeshare studies that require accurate trip distances. Currently, Capital Bikeshare uses an average speed of roughly 7.5 MPH to estimate distances.
Trip Averages |
Casual |
Member |
All Users |
|||
Mean | Median | Mean | Median | Mean | Median | |
Time (minutes) | 39.72 | 25.77 | 12.50 | 10.02 | 18.97 | 12.07 |
Speed (MPH) | 4.71 | 2.32 | 7.85 | 8.38 | 6.29 | 7.83 |
Distance (miles) | 3.12 | 5.50 | 1.64 | 1.36 | 1.99 | 1.53 |
Difference | Difference | Difference | ||||
Direct distance | 1.12 | 177.59% | 1.21 | 35.10% | 1.19 | 67.07% |
The below heat maps display miles ridden for each user type. Not surprisingly, casual-user activity follows the shape of the National Mall. It is also clear that some casual users ride to Georgetown, Arlington Cemetery, and Crystal City. (In upcoming articles I’ll examine where casual users stop their bikes to check out the sights, and also determine the most popular streets taken for both users.)
Member activity centered further north, in an area bounded roughly by 13th, 17th, K, and R Streets NW. It is clear from the data that member riders tend to stay clear of the National Mall and focus more in the mixed-use neighborhoods of Dupont and Logan circles. Also evident is a large percentage of member users coming from Arlington and Alexandria into the city.
Looking at individual streets, the most popular segments for casual riders are the sidewalks within the Mall and the roads that border it (see below). The Mount Vernon Trail in Virginia was also popular among casual riders.
Segments from casual trips. Click to enlarge.
The most popular segments for member riders were 14th, 15th, 18th, and R Streets NW. Also popular was Pennsylvania Avenue in front of the White House and the Key Bridge, linking Georgetown with Arlington, Va. Most of these popular segments include bicycle infrastructure such as bike lanes or protected lanes.
Segments from member trips. Click to enlarge.
Look for my next article here at Mobility Lab. It will examine specific bicycling infrastructure usage in more detail.
Maps and video by the author.