Studying Lyon's Vélo'V: A Statistical Cyclic Model

Lyon's community bicycle program called Velo'v is a major initiative in shared public transportation, in activity since May 2005. It is studied here at a global level, to assess the evolution with time of the number of hired bikes. Based on the entire Velo'v data set, up to December 2007, a statistical model is proposed to describe the daily and weekly patterns in a cyclostationary manner, jointly with the non-stationary evolutions over larger time-scales larger. Combining this model with linear statistical regression, a procedure is developed for the prediction of the number of bikes hired per hour. This prediction method involves several explanation factors such as the number of subscribed users, the time in the week, the occurrence of holidays or strikes, and weather parameters (temperature, volume of rain). The conclusion is that, for most days, the observation of the number of actually hired bicyles is satisfyingly explained and predicted by the model proposed here.