Analyzing Cyclist Behavior with Velo'v Bike Share

Earlier this month, an interesting preliminary study was released from France detailing cyclist behavior with data collected from the Velo’v bike share system in Lyon, France. The Velo’v bike sharing system is a very extensive system with approximately 4000 bicycles available at 343 stations around the city of Lyon. For the most part, the Velo’v system has bike stations located within every 300 meter buffer zone throughout the city. The extensive coverage, reasonable rental and subscription fees, and high-tech access and locking systems have made Velo’v one of the most famous bike sharing systems in the world, and one after which many other systems are modeled.

Velo'v MapThe extensive coverage of Velo’v has also provided a unique opportunity to collect information and data about the behavior of bike share users. The advertising company JC Decaux, which operates Velo’v, has been collecting data about the use of their bicycles since the system was implemented in May of 2005. JC Decaux provided a data set of Velo’v bicycle use from May 2005 to December 2007 to a group of researchers in France. The data set yielded some potentially interesting results about the behaviors of urban cyclists, which could in turn help urban planners best decide how and where to add bicycle infrastructure.

The study looked at 11.6 million trips over the course of the two and half year period. Each record in the dataset gives information about the location and time of the beginning and end of each trip, as well as the trip distance. The data set indicated that the average trip distance is 2.49 km (1.55 mi) and that each trip takes an average of 14.7 minutes. The Physics arXiv Blog summarizes many of the other results, so I will not repeat them all here, but in general, the data set and the initial paper show some very interesting trends in the use of Velo’v bicycles, especially trends that show peak Velo’v use throughout the week, as well as peak use hours throughout the days of the week.

Bike use graphThe data set also revealed some rather odd results, including an increase in average speeds on Wednesdays, an increase in speeds during rush hour that outpace average car speeds during rush hour, and shorter distances for bicyclists than car drivers between two points indicating the use of shortcuts.

Notably, Lyon has few bike lanes, so Jensen et. al. who wrote the initial paper suggest that Velo’v users are riding on sidewalks, the wrong way, and more. Jensen et. al. also suggest that the higher speeds on Wednesdays are a result of higher proportion of men riding on Wednesdays, as women tend to stay at home to watch children on that particular day of the week in France.

In general, I think there is a lot of opportunity to learn about cyclist behavior from the Velo’v dataset, despite some of the problematic assumptions made by Jensen et. al. about gender, shortcuts, and more.

Additionally, it is important to note (and I can’t seem to find this in the Jensen et. al. paper) that this study is not really representative of urban bicyclist behavior in Lyon (or elsewhere, for that matter), as it does not have any information about people who own their own bicycles and do not use the Velo’v bikes. Nonetheless, it is encouraging to see that this type of research is being conducted and hopefully there is more to come, either from other cities, bike share systems, and other types of studies. There have been calls of late in the U.S. for more hard numbers and data on bicycle use. But it is also important to realize that drawing conclusions from narrow datasets (such as the Velo’v dataset) can be problematic if they are used to make broad, sweeping generalizations about bicyclist behavior.

Post navigation