An Intelligent Bike Sharing Prediction System



Bike-Sharing Systems offer bikes-on-demand to users at an affordable price, improving urban mobility. It is currently the fastest growing mode of transportation in the world, with an aggregated growth rate of around 45% since 2007. In the summer of 2014 there were more than 600 cities with a bike-sharing program. Nowadays one of the main problems and challenges in the manage- ment of Bike-Sharing Systems is to assure that users will be able to find available bicycles or park them in a station at any time. However, users behaviour causes bicycles to be asymmetrically distributed. Therefore a Rebalancing System is needed to maintain the adequate number of bikes at each station in order to satisfy the demand. Rebalancing is a costly operation in terms of logistics and its inefficiency represents a main cause of dissatisfaction within customers. The design of algorithms to manage rebalancing helping customers and operators to know where to drop or pick bicycles accurately and efficiently is a very important step in terms of sustainability of the system. In this paper, we present an intelligent prediction system that represents an additional value to the project Smart BikeEmotion developed by Ubiwhere: Suiting the Future Lda aiming to predict users needs and learn which stations need to be rebalanced depending of the current location, weather, time and topology of the city. This rebalancing algorithm is compared to current real time ones used in today’s Bike-Sharing Systems. Its advantages, limitations, and lessons learned from its implementation are analysed and possible improvements are addressed in this paper.


Bike-sharing, RebalancingProblem


Bike-sharing systems


18th Euro Working Group on Transportation, EWGT 2015, 14-16 July 2015, Delft, The Netherlands, July 2015

Cited by

No citations found