Posted by: jdantos | January 18, 2012

Capital Bikeshare Data, Part 2

Part 2 in a (now?) series of posts analyzing Capital Bikeshare usage data. This post focuses on system-level usage by trip duration. See parts 1, 3, and 4, 5, 6, 7 (maps of travel patterns), 8, 9 (weather).

When Capital Bikeshare riders get on those big red bikes, how long do they keep the bike out? Are most trips for a 5-minute jaunt to Dunkin’ Donuts where you’re too lazy or hurried to walk? Or are these longer trips?

Here’s a frequency distribution of all 1.36 million trips, grouped into buckets of 2 minutes each, casual and registered users alike.  For trips lasting more than 60 minutes, way out on the right of the graph, I started grouping rides in larger chunks of half-hours and hours. (Zero and below is bad data).

Capital Bikeshare Trips by Trip Duration by User Type

Capital Bikeshare Trips by Trip Duration by User Type (click for larger)

The single most popular trip duration for all users is 4-6 minutes, followed by, surprisingly, 10-12 minutes.  That would suggest that alot of users are riding for short (like, 1 mile) trips, but also hints that there may be two sub-markets here (would need to break this down by geography and time of day).

The vast majority of trips (88%) are less than 30 minutes.  At first, I figured that meant that registered users are just savvy and are watching the clock to avoid fees.  But, there isn’t a steep dropoff right at 28 or 30 minutes.  In fact, there have been more rides between 30-32 minutes (barely incurring fees), than 28-30 minutes (just under the limit of free time).  Instead, maybe 20 or 25 minutes (roughly 2, maybe 3 miles?) is just the maximum amount that average users are willing to ride.  Anything longer, and they’ll choose another mode.  Or, it could be that with 2 or 3 miles you can just get to most things in the D.C. and Arlington core.  Or both. What do you think?

Also, what’s up with the casual users? The distribution of their trip lengths is really different. Let’s look at that same graph, but withe casual users only:

Capital Bikeshare Trips by Trip Duration,  Casual Users Only

Capital Bikeshare Trips by Trip Duration, Casual Users Only (click for larger)

The distribution is much more even – only 59% of trips are under the 30-minute threshold.  For casual users, 40% of all trips exceed the 30 minutes and incur fees. And, there’s a big spike of hour-plus usage for casual users.  23% of their trips are 60 minutes or more.  It seems that a good chunk of casual users are also using the bikes for 1, 2, even 3 hours.  Are these tourists riding around the monuments all afternoon, perhaps caught near the Tidal Basin and far from a station?  Are they inelastic to (or unaware of?) the fees that start rolling in after 30 minutes?  Will this tail tend to drop off as new stations are installed on the Mall?

Does trip duration change by season? Does winter’s bitter cold keep rides short? Here’s the same frequency distribution, but broken down by season (I couldn’t fit user type on there too).

Capital Bikeshare Trips by Trip Duration by Season

Capital Bikeshare Trips, by Trip Duration, by Season. (click for larger)

Turns out, the distribution of trip durations stays pretty constant by season.  As use declines in the winter, it declines across all trip lengths, and vice versa in warmer weather. There’s a pretty consistent spike of use around the 5-minute mark, and then again in the 10-12 minute mark. And, the “tail” of hour-plus rentals is biggest in the summer and spring, but still persists a little in the winter too.

How is this useful? While it’s hard (but possible) to make inferences about where riders go or what path they take (although OObrien is doing a great job), knowing the length of the rental might help estimate distance traveled. Especially in cases where return station is same as rental station, or where imputed speed is so low that there must’ve been a pitstop or a loop.

More to come next week, stay tuned.  I’d like to explore station-to-station link data, usage vs. capacity, and the effects of weather. One thing that would help is a way to group the stations into zones or neighborhoods – reducing the number of station-to-station pairs might help visualize flows.  Any good ideas?

About these ads

Responses

  1. I wondered if the 30 minute limit contributed to people running red lights, but this data seems to suggest that this is not an issue.

  2. […] two posts, one that deals with various facets of Bikeshare use, and another that discusses CaBi in terms of trip duration. Thanks for helping us make sense of this information, […]

  3. I think that you’re mis-analyzing a part of these data. In your histograms, you change your bin size after one hour. It’s possible that there’s an actual spike but I bet it’s just an artifact of this shift.
    I tried to clean up the CaBi data myself but ran into some issues. Can I see/have your data so I can recreate the histograms to (possibly) illustrate what you refer to as the long tail?
    Cheers!

    • David – yeah, I changed the bin size after 60 minutes because otherwise, continuing in 2-minute increments would’ve made the graphs impossibly wide. Instead, I made the judgment call that trips over an hour were probably a different beast than those under and hour, and binned accordingly. So it makes a blip on the graph, perhaps unfair to the eye, but sure enough, the majority of hour-plus trips are by casual users.

      I’m happy to give you my data – it’s just the 5 CSV files from CaBi,merged into one Access table. Is Access 2010 okay?

      • Perfect!
        (You’ve got my email via your blog?)

  4. David F-H
    If it helps, I merged the 4 files (2010 data only) into a single XLSX file -scrubbing all rides <2min.
    Only problem is that the file is 75MB, (so it helps if you're working with 64bit Excel, 6GB+ of RAM and a good processor)

    • Cool!
      I’d love it!
      so, here’s me putting my email address on the wider inter-webs.
      friedlad at gmail dot com.
      Thanks!

      • David
        Sorry for the very delayed response. If you see this, I am uploading the CaBi data now and will email you details shortly

  5. […] certain areas are more in balance than others. If you’re bored, nerdy, or both, read Part 1, 2, or […]

  6. I do not mean to nitpick, but I believe CABI does not charge until 31 minutes. 30 minutes and 59 seconds does not incur usage fee’s. The spot cycle rental timer default setting goes off at 25 minutes into the ride. Perhaps consolidating the data between 25 and 31 minutes will give a better idea of the “avoid fee’s is my top priority” crowd. Of course a lot of people have their own internal clock and set commute, knowing when and where to re-dock to avoid usage fee’s.

  7. […] post quickly looks at how much each bike gets used and churned through the system. See parts 1, 2, and […]

  8. […] Bikeshare usage data. This post focuses on system-level usage by a few dimensions. Check out parts 2, 3, and […]

  9. […] call this part 5, even though it’s kind of a postscript to part 4. See parts 1, 2, and […]

  10. […] Capital Bikeshare data. This one focuses on net flows of bikes across the system. Parts 1, 2, 3, 4, and […]

  11. […] crowd-sourced Capital Bikeshare data. This one maps a bunch of data by station. See also parts 1, 2, 3, 4, 5, and […]

  12. […] simple correlation between temperature and usage. If you’re really bored, see also parts 1, 2, 3, 4, 5, 6,7, and […]

  13. […] a dozen pedal strokes and you’re at the next block already. This may be the genius behind Bikeshare’s short trips. Most utilitarian city bike riding is not hardcore, not masochistic, and not a crazy feat of […]


Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

Categories

Follow

Get every new post delivered to your Inbox.

%d bloggers like this: