This bike share system is designed for short trips, between bike stations. Of the 144,000 trips in the data set, the median trip length was only nine minutes. And, about 96% of trips were less than one hour long. This makes sense since there is a late fee charged if a bike is not returned to a station within half an hour. Of course, one could go on longer rides without incurring a late fee by hopping between stations, as long as each hop is shorter than a half hour; however, the data set does not provide enough information to infer whether a trip was part of a multi-leg ride.
|The median trip length was nine minutes.|
About 4% of the trips (5696) were longer than one hour and, thus, incurred a late fee. In fact, the average length of these long trips was about 4.5 hours. Since a late charge of $7 applies to every half hour after one hour, this amounts to $49 in late fees, for a total of about $280k in late fees collected overall. Now that's a great revenue stream! :)
Why are these bikes returned late? Did the rider not know about the late fee? Did they get lost? I do not know, but a vast majority of these trips, 94% (5379), were taken by short-term customers (customers with 1- or 3-day memberships), as opposed to subscribers (who hold annual memberships), even though short-term customers are represented in only 21% of all trips.
The top three stations at which trips started were all in downtown San Francisco:
- San Francisco Caltrain (Townsend at 4th) : 9838
- Harry Bridges Plaza (Ferry Building) : 7343
- Embarcadero at Sansome : 6545
However, if broken down by membership type, the most popular stations were:
- Among short-term customers (1- or 3-day memberships): Embarcadero at Sansome
- Among subscribers (annual membership): San Francisco Caltrain (Townsend at 4th)
Day of Week
The weekly patterns also vary by membership type. Short-term customers tend to take more trips over the weekend, while subscribers tend to ride on week days:
|Customers (with 1- or 3-day memberships) preferred weekends.|
|Subscribers (with annual memberships) preferred weekdays.|
The above analysis was only on trip data. The data set also contained weather data, bike station information, and station rebalancing data. If you are interested in taking a look, check out the data from Bay Area Bike Share and the R code used in my analysis.