Capital Bikeshare’s trip history data for the 4th quarter of 2012 has been posted, and with it I’ve made a new version of the CaBi Trip Visualizer. (See last quarter’s version at A Closer Look at Bikeshare Data.)
For this version, I’ve strengthened the features that let you examine clusters of stations. When you select a single station, the program draws arrows to stations that got a significant amount of traffic from that station. The thickness of the arrows is scaled so that the station with the most traffic gets a line 10 pixels thick. Anything less than a tenth of that level isn’t shown.
Since this is the first quarter for which Alexandria has been in full operation, let’s start there. By looking at the seven stations as a cluster, you can see how most traffic is going to or from the two Metro stations, King Street and Braddock Road. This seems to indicate the bikes are being used to solve the “last mile problem,” which is a prime benefit touted by many bikesharing advocates.
Also, note it doesn’t show any trips going outside of the cluster, meaning most riders use the system to stay within Alexandria. There are of course riders who leave the area, but the numbers are too small to show up on our map. Of the 4,139 bikes checked out in Alexandria, 91 percent were returned to Alexandria. Five percent went to Crystal City, 3 percent to D.C., and fewer than 1 percent went to the Wilson Boulevard area of Arlington.
How heavily do tourists use the new system in Alexandria? One way to judge this is to look at the number of “loop trips” taken, meaning trips that begin and end at the same station. Some of these may be round-trip errands, but I suspect that many are simply joy rides taken by tourists. Nine percent of Alexandria’s trips are loop trips, compared with the system-wide average of 3.5 percent.
We can also look at who’s making the trips. Eighteen percent of all trips beginning in Alexandria are taken by casual riders (who have temporary one- or five-day memberships). For the system overall, the percentage is only 14 percent. So, though the bulk of trips are taken by registered users (who have monthly or annual memberships), it’s not as dominated by registered users as the overall system is. (Note the Visualizer doesn’t differentiate between rider types. I had to make custom queries of the database.)
Georgetown’s cluster patterns are very different. You can see that most trips are going in or out of Georgetown, and not within. Of the 12,500 bikes checked out in Georgetown, only 11 percent were returned to Georgetown. Eighty percent of the trips went elsewhere in DC, 4 percent went to Rosslyn, and the remaining 5 percent went elsewhere in Virginia. To me, the heavy traffic documents a need for improved bikeshare connections east of Georgetown, both on-street access and improving the trail in Rose Park.
Five percent of bikes checked out of Georgetown’s four stations are taken on loop trips. Twenty-nine percent of Georgetown’s trips are made by casual riders.
H Street, NE
Looking at the cluster of five stations along H Street, NE, it may seem that they are being used primarily to go to and from Union Station. And while Union Station is clearly the most popular end-point, those thin lines going elsewhere add up to a greater amount. Of the 12,349 trips that began on H Street, 16 percent went to Union Station, while another 16 percent went elsewhere on H Street. Sixthy-eight percent went elsewhere in D.C. (An insignificant 0.1 percent went to Virginia.)
Three percent of bikes checked out of H Street’s five stations are taken on loop trips. Only 8 percent of H Street’s trips are made by casual riders. These low numbers tell me that H Street is used more heavily by serious CaBi members, and hasn’t yet been fully discovered by the tourists.
Washington D.C. zoomed out
Zooming out, let’s take a quick look at the cluster of stations in DC. It’s interesting to see Capitol Hill emerge as a network that’s separate from the one that spans Dupont Circle and Shaw. For more stats on the 4th quarter’s data, see Capital Bikeshare’s 4th Quarter of 2012.
It’s fun to see the numbers used to create a display that tells a story about how the system is being used. What other question do you have about the system that could be answered by visualizing the data?
For some additional user tips, here’s how you can select a cluster of stations. I’ve built in my own choices for 13 clusters, listed below. You can pick these groups by finding them at the bottom of the drop-down menu (listed after single stations), or you can use shortcut keys:
- A: Arlington
- C: Crystal City
- D: D.C.
- E: East of the river
- F: Fourteenth Street
- G: Georgetown
- H: H Street
- M: The Mall
- N: Nationals stadium
- O: Old Town Alexandria
- U: Connecticut Avenue
- V: Virginia
- W: Wilson Corridor
Or, create a custom cluster by picking which stations to include. Pan and zoom the map so that only your stations are visible, then press the “1” key to draw the cluster’s network.
For clusters, lines are drawn for traffic to/from any of the included stations, and scaled to the maximum station-pair found.