A Closer Look at Bikeshare Data

avatar
Michael is senior tech advisor for Mobility Lab. He specializes in photography and cool transportation visualizations, and he leads Transportation Techies DC.
January 8, 2013

Cabi Flow Analyzer

Every quarter, a new set of data gets added to Capital Bikeshare’s repository of trip-history data. And as the Washington D.C. region’s bikesharing system grows more popular, the data sets grow in size (from 15 megabytes and 117,972 trips in the 4th quarter of 2010 to 68 megabytes and 637,531 trips in the 3rd quarter of 2012).

How best to sift through the mounds of data and find meaning?

Cabi FlowThe “CaBi Trip Visualizer” is a new tool I developed that lets you analyze 2012 3rd-quarter Capital Bikeshare data, using arrows along a map to illustrate where riders are going to and coming from. The thickness and opacity of the arrow reflects the relative ridership levels. Hovering over an arrow presents a display of the numeric values for that segment.

You can try out the tool at http://mvjantzen.com/cabi/trips3q2012.html.

For each station you select, the most-common route is drawn with an arrow 10-pixels thick. Segments to other stations that are less than 10 percent of that value are not included in the map (since the lines would be less than 1 pixel). This also prevents outlier trips from forcing the boundaries of the map’s view to be too large. (You can still zoom and pan within the map.) Be sure to play with the multiple-station options, at the bottom of the drop-down menu, such as looking at travel patterns for stations on the Mall.

Notice any insightful patterns? Ideas for how to improve the tool? Let us know what you discover in the comments below. (The JavaScript behind the scenes was modified from a bubble map made for the 2011 4th-quarter data, which you can see at Looking at CaBi Stats with a Bubble Map.)

More Number-Crunching

  • Hungry for more statistics from the new data? The average distance between end-points was 1,858 meters (1.2 miles). This excludes trips that began and ended at the same station (as happens in almost 5 percent of all trips). And of course the trips were certainly longer than the distance calculated as the crow flies.
  • 80 percent of the third-quarter’s trips were made by registered users (those who sign up for a month or a year). Casual users (who made 20 percent of the trips) sign up for 1 or 3 days.
  • The busiest station by far is the one on Massachusetts Avenue, just west of Dupont Circle (measuring check-outs and ignoring check-ins). In fact, 2.8 percent* of all bikeshare trips begin at the Dupont Circle station. The second-busiest station is the one at 15th & P streets NW (1.9 percent of all trips). And the third-busiest station is at 14th & V streets NW, outside of the Reeves Center (1.6 percent of all trips).
  • The casual users prefer the stations closer to the Mall. Their favorite station for checking out a bike is at Jefferson Drive & 14th Street SW. That station is used to begin 5.8 percent of all casual trips. It’s followed by the one at 19th Street & Constitution Avenue NW, which gets 4.1 percent of all casual trips, and then in third place we’re back to Massachusetts Avenue & Dupont Circle NW, where 2.9 percent of all casual trips begin.

*Keep in mind that with 190 stations, each station represents 0.53 percent of the total.

{ 11 comments… read them below or add one }

avatar SonT January 8, 2013 at 4:09 PM

Hi, could you share what tools, programming languages, APIs, etc. did you use to create this data visualization? I’m guessing it involved some Google Maps?

It’s brilliant! I’ve long wondered where are rides coming from/going to and now I can see it for myself!

I hope Capital Bikeshare can use the insights from this data viz to better address the “unbalancedness” issues between stations.

THANK YOU!!!!

Reply

avatar Mr Michael Schade January 8, 2013 at 5:17 PM

You can see the JavaScript code by viewing the source of the page. Yes, it does in fact use the Google Maps API (v3, for JavaScript). Lots more examples of playing with the Google Maps API on my blog: http://www.mvjantzen.com/blog/?tag=googlemaps Glad you liked the CaBi Trip Visualizer!

Reply

avatar Sean Robertson January 8, 2013 at 10:33 PM

What I’d really love to see is a way to do this with one’s own personal trip history. Bikeshare is currently redesigning the way member trip histories are displayed. Wouldn’t it be fantastic if we could persuade them to provide the data as a .csv file? If we can, perhaps those files could be uploaded to create maps like yours? I’ve always wondered what my personal map would look like!

Reply

avatar Francis Chen January 9, 2013 at 1:47 AM

Hey, have you considered doing something similar with the Hangzhou Bike-sharing System in China? While currently the biggest bikesharing system in the world, as far as I know of The HZ Bikesharing system don’t have a real time system showing where bikes are picked up, and then dropped off.

I was able to obtain the XY coordinates of all HZ Bikesharing stations from the HZ Bikeshare Website. Here they are in Map and Table form:
https://www.google.com/fusiontables/DataSource?snapid=S374515Xvjy

Here’s a picture:
http://www.flickr.com/photos/fncischen/6696553021/in/set-72157628610445357

My Chinese isn’t the most fluent, but if you are looking for particular terms or data, let me know and I’ll attempt to find it. Just keep in mind that transportation data infrastructure is quite tricky to find in China specifically

Reply

avatar Francis Chen January 9, 2013 at 1:51 AM

Oh yes, I do sort of feel guilty to send your work to the HZ bikesharing people haha.

Reply

avatar Mr Michael Schade January 9, 2013 at 2:44 PM

I’d love to do the same analysis for other cities. The Hangzhou system would be interesting, but I can’t navigate the Chinese web site http://www.hzzxc.com.cn/ The list of stations is not enough; I need a plain-text (such as CSV) dump of the usage data, such as the files CaBi lists at the bottom of https://www.capitalbikeshare.com/trip-history-data

Reply

avatar Avocado December 9, 2013 at 9:41 AM

Can you update this map with the new stations? I’d love to see where people in Rockville, Takoma, Silver Spring are riding to — if they are riding to DC or only using the bikes for trips between nearby stations. Thanks for putting this great map online!

Reply

avatar Mr Michael Schade December 9, 2013 at 1:34 PM

CaBi releases their trip-history data quarterly, so we have to wait for the new year before we get data showing trip patterns for the new stations in Montgomery County. Once the data comes out, we look forward to updating the tools with the new data!

Reply

avatar Michæl Schade January 26, 2014 at 8:51 PM

The 2013 data has been added. See http://mvjantzen.com/tools/visualizer/?system=cabi Some analysis of the new Montgomery County stations: http://www.mvjantzen.com/blog/?p=4544 …more to come!

Reply

avatar Bill Schneider June 7, 2014 at 11:39 PM

This is awesome. Wondering how hard it would be to add a time of day component (maybe animated?) to show how patterns change over the course of a day. I try to use the stations near the U Street Metro station (3 docks near the Green line) to complete the last mile of my commute to the West End, and it’s very hard to find a bike there in nice weather so I’m curious where the bikes all end up.

From slicing some of the data in Excel, I noticed the patterns are different during morning rush hour, and many of the bikes end up very close to another Metro station on a different line downtown. 13th St and New York Ave NW (Metro Center), 20th St and L St NW (Farragut), and 15th and K St NW (McPherson Sq) are all popular destinations from the U Street Metro area during morning rush (7am-9am weekdays) but not popular at other times.

I suspect many of those riders during the morning rush are getting off the Green Line from the suburbs and using CaBi to avoid a longer ride and a transfer, with their final destinations close to a Metro stop on a different line.

Reply

avatar Michæl Schade June 9, 2014 at 11:13 AM

Good idea! It would be interesting to see someone analyze that. Maybe at the next Transportation Techies CaBiHackNight?

Reply

Leave a Comment

{ 7 trackbacks }

Previous post:

Next post: