The Bay Area Bike Share system allows users to rent bicycles for short journeys between stations throughout the city. Users can be annual members or short term (1 or 3 days). The system is completely automated for users.
There are 69 stations across 5 cities in the Bike Share system, with an average of 17 docks per station.
About 50% of the stations and docks are located in San Francisco, but it makes up 90% of the system use.
The data we analyze here came from rides between August 29, 2013 and February 28, 2014.
In those 185 days there were 144,015 rides, averaging about 20 minutes per trip.
Total riding time:
49,241½ hours, or 5 years, 32 weeks, 1 day, 12 hours, 27 minutes, and 4 seconds.
This chart makes it pretty clear that ridership drops on weekends. Check the box to highlight Saturdays and Sundays.
917 rides are made on average each weekday
442 rides are made on average each weekend
|Busiest day||Calmest day|
|1,264 rides||81 rides|
Almost 80% of riders have an annual subscription. The remaining 20% of riders were customers who purchased a 24-hour or 3-day pass.
We saw already that users are likely to be annual subscribers, and system use drops off on the weekends. Let's group the rides by weekday and see how subscribers' use compares to customers':
Weekday riders are overwhelmingly subscribers, and ridership among subscribers falls on weekends so that rides by customers just manage to outnumber them.
And now we'll group rides by time of day:
Among subscribers we see spikes in use at 8am and 5pm with another small bump at 12 noon. These users must be riding a bike to get to work, to go to lunch, and to head home.
Customers hourly usage seems to fall along a bell-shaped distribution peaking at two in the afternoon. There doesn't seem to be a lunchtime rise among customer use. These users must be riding around throughout the daytime at their leisure.
From these usage behaviors, it would be fair to characterize the two groups of subscribers and customers as commuters and tourists, respectively.
The Bike Share system is intended to be used for short rides: trips under a half hour do not incur any additional charges. Do riders use the system in the intended way?
Yes they do. The chart on the right shows that the most common ride length is 5 to 10 minutes. Subscribers are clearly savvy to the price structuring. Very few rides longer than a half hour are taken by subscribers. For the most part, customers are savvy as well. Their trips last a little longer on average, but mostly less than 30 minutes. There's a bump in rides that last longer than an hour, let's examine those a little closer:
By the current pricing structure, a customer purchasing a 24 hour pass and taking a ride that lasts 2 hours and 59 minutes would pay in total $41. A ride of 2 hours and 29 minutes would only be $34. There are several bike rental companies in San Francisco which offer 3-hour rentals for around $32. Customers taking trips less than three hours likely did not understand the bike share system and incurred unwanted overtime fees, or they decided the convenience of the automated system was worth the premium paid over other rental offerings.
Rides lasting longer than an hour are most commonly 1 to 2 hours long, and almost always taken by customers. This could be due to confusion over the nature of the "24-hour pass" or one of many other factors including theft, forgetfulness, or getting lost.
We saw earlier that 90% of Bike Share rides took place in San Francisco. Let's examine the average numbers for each city on a day-to-day basis:
|System Wide||San Jose||Redwood City||Mountain View||Palo Alto||San Francisco|
Hmm. Redwood City and Palo Alto don't see much daily use, but not much interesting here. Let's look at a chart. Let's look at an interactive chart.
Here we see total daily rides for each category of user plotted across the entire timespan, similar to the chart at the top of the page. Use the radio buttons on top to examine different cities and the checkboxes below to highlight days with a factor that might have influenced riders.
If our generalization of users as commuters and tourists were true, we would predict use among subscribers to drop off when commuters don't go to work and increase when tourists are in town. With the overlays we can see this correlation holds true. Rides by subscribers in all cities decreased on weekends, over Thanksgiving, and between Christmas and New Years. Rides by customers increased noticeably on weekends, holidays, and during the America's Cup finals.
Rain seems to be a large deterrent to bike use; on rainy days system use falls almost as predictably as on weekends or holidays. Temperature, however does not seem to have a major influence on riders. Bay area temperatures being notoriously moderate, the lack of correlation is not surprising.
We can't draw any conclusions about the influence of home games by the 49ers, Giants, or Sharks. Heading to a Giants game would seem to be an ideal use of the Bike Share, but during this period the only Giants home games overlapped with Americas Cup races. There are a few dates where a Sharks game correlated with a slight rise above average usage, but such deviation from average is normal throughout this time period.
Interestingly, both the federal government shutdown of October and the two BART worker strikes of October and December did not seem to have a major influence. Rider numbers during the strikes remained similar to numbers from the weeks before and after the strikes. This poses an interesting question, do BART commuters use the Bike Share System? To begin looking for an answer, let's look at the numbers for rides to and from individual stations.
|San Francisco Caltrain (Townsend at 4th)||9,838|
|Harry Bridges Plaza (Ferry Building)||7,343|
|Embarcadero at Sansome||6,545|
|Market at Sansome||5,922|
|Temp. Transbay Term. (Howard at Beale)||5,113|
|Market at 4th||5,030|
|2nd at Townsend||4,987|
|San Francisco Caltrain 2 (330 Townsend)||4,976|
|Steuart at Market||4,913|
|Townsend at 7th||4,493|
|San Francisco Caltrain (Townsend at 4th)||11,637|
|Embarcadero at Sansome||7,590|
|Harry Bridges Plaza (Ferry Building)||7,475|
|Market at Sansome||6,238|
|2nd at Townsend||5,655|
|San Francisco Caltrain 2 (330 Townsend)||5,112|
|Market at 4th||5,109|
|Steuart at Market||5,080|
|Townsend at 7th||5,073|
|2nd at South Park||4,431|
|Most popular starting point||Most popular destination|
|9,838 rides||11,637 rides|
Here we see the top 10 stations in the entire system to start or end a ride. Interestingly, the stations near Caltrain and the Ferry building top the list. Both Harry Bridges Plaza and Steuart at Market are close to the Ferry Building. San Francisco Caltrain 1 and 2 obviously serve the Caltrain line. The station at Embarcadero at Sansome is next to Pier 27, where the America's Cup Pavilion was located.
The Bike Share stations on the top ten lists within walking distance of a BART stop are Market at Sansome, Market at 4th, and Steuart at Market. Keep in mind that all these stops are also within close proximity to many points of interest in downtown San Francisco.
Of course, the rest of the stations serving BART commuters could all be just outside the top ten. Let's take a look at the list of rides for all stations. BART stations are identified with blue text, Caltrain stations with red text.
While the BART stations are indeed all among the busiest stations, they are all in San Francisco, and their numbers don't set them apart from any other station within San Francisco. These stations are centrally located to a number of attractions in downtown San Francisco, so we would expect them to be highly trafficked. Bike Share Riders docking at these stations are not necessarily transferring to or from BART. If there was a large contingency of BART commuters who used the Bike Share, we would expect to see rides to and from BART stations in numbers setting them above other destinations. With what we've seen here, it seems safe to conclude that BART commuters do not make up a large proportion of Bike Share riders.
Is there any more evidence to back up this assumption? Let's take a look at the rides at stations serving different forms of commuting to and from San Francisco:
|BART||Caltrain||Ferry Building||Transbay Terminal|
These numbers are somewhat misleading since there are four stations within walking distance of a BART station, but only two nearby the Caltrain and Ferry building. Instead, let's look at average rides to/from these stations:
|BART||Caltrain||Ferry Building||Transbay Terminal||All SF|
Bike Share stations serving BART do not seem to be more trafficked than other stations in San Francisco. Of course, we would be making a massive mistake if we were to assume that all rides to or from these stations started or continued with trips on the related commuter system. The Ferry Building is a major tourism destination, every BART stop is within walking distance to numerous restaurants, museums, business, MUNI rail and bus stops. The Caltrain station, on the other hand, is only within walking distance of a handful of other points of interest. When we compare the numbers for the Caltrain station to the average for all stations in San Francisco, it becomes clear that Caltrain commuters are making great use of the bike share system.
Each trip within Bike Share is recorded by starting station and ending station, so in addition to examining the popularity of stations, we can examine the popularity of individual routes. In the heatmap below, starting stations are listed in rows and the ending stations in columns. The total number of rides from least to most along each route is indicated by the cell color, lightest to darkest.
On the heatmap diagram of systemwide rides we see that activity tends to be grouped into squares. These are trips that took place within city boundaries, and we notice that not many riders go beyond their starting city. An exception to this are trips between Mountain View and Palo Alto, between which 191 rides went from one city to the other.
Looking at the station heatmap for San Francisco, we note that riders leaving from the most popular station, SF Caltrain, disperse throughout the system. Riders heading to SF Caltrain similarly tend to come from throughout the system.
Two popular routes which seem to make up a round trip are Townsend at 7th to SF Caltrain 2 and vice-versa. Near to Townsend at 7th is an Expo center and at handful of influential tech companies including Adobe, Heroku, Citrix, Advent, and Zynga. This could reflect a corporate membership program popular among commuting employees, a conference well-attended by peninsula-dwellers, or more simply, a dearth of other forms of transportation to the area.
The most ridden route, Harry Bridges Plaza (Ferry Building) to Embarcadero at Sansome, does not have a return route with as great numbers, indicating that riders tended to ride to Embarcadero at Sansome and continue their journey elsewhere rather than return to the Ferry Building. Embarcadero at Sansome is the northernmost station along the Embarcadero, closest to tourism-heavy Pier 39 and Fisherman's Wharf. The bike path heading north along the Embarcadero is also much more bicycle-friendly than the southern route.
|Harry Bridges Plaza (Ferry Building)||Embarcadero at Sansome||1,330|
|Townsend at 7th||San Francisco Caltrain (Townsend at 4th)||1,322|
|San Francisco Caltrain 2 (330 Townsend)||Townsend at 7th||1,116|
|Market at Sansome||2nd at South Park||866|
|Embarcadero at Sansome||Steuart at Market||811|
|2nd at South Park||Market at Sansome||798|
|San Francisco Caltrain (Townsend at 4th)||Harry Bridges Plaza (Ferry Building)||782|
|2nd at Townsend||Harry Bridges Plaza (Ferry Building)||757|
|Steuart at Market||Embarcadero at Sansome||717|
|Harry Bridges Plaza (Ferry Building)||2nd at Townsend||710|
We saw here that Bike Share is used mostly in San Francisco, by commuters, when it isn't raining, for rides under 15 minutes. It looks like BART commuters don't use the system to the extent that Caltrain commuters do. On weekends and holidays, visitors to San Francisco use the system to ride around town.
One dimension not explored in this analysis was popularity of each station among subscribers vs customers. With what we saw of the behavior of customers vs subscribers we could identify stations more popular with tourists or with commuters and potentially identify areas of the city with demand for future stations.
One recommendation that can be made is for station growth into SOMA. Users are already shown to be commuters heading to work, and SOMA is one area of San Francisco with a high density of businesses, close proximity to the existing stations, and currently lacking any Bike Share stations.
After submitting this entry to the bike share contest, I kept thinking to myself, "It doesn't make sense that BART riders don't make up a big group of Bike Share users. How else could I show where the Subscribers come from?" It turns out that information was in the original dataset all along, so I made a map of subscribers by zip code.
Which shows that many Riders are Subscribers who live in the East Bay, near a BART station. Clearly, some more analysis is in order to figure out why the BART Bike Share stations don't stand out among other San Francisco stations.