Bike sharing systems continue gaining worldwide popularity, as they offer benefits on various levels, from society to environment. However, the system tends to be unbalanced along time. Reasonably accurate demand prediction is key to effective redistribution; however, it is has received only little attention in the literature. In this paper, we focus on predicting the hourly demand for demand rentals and returns at each of the stations. The proposed model uses temporal and weather features to predict demand mean and variance. It first extracts the main traffic behaviors from the stations. These simplified behaviors are then predicted and used to build a global station-level prediction using machine learning and statistical inference techniques. We then focus on determining prediction intervals, which can be directly used by the bike sharing companies for their online rebalancing. Our models are validated on a two-year period of real data from Bixi Montréal. A worst-case analysis suggests that the intervals generated by our models may decrease unsatisfied demands when compared to the current methodology employed in practice.
Bienvenue à tous!