A Markov-chain based model for a bike-sharing system

Abstract

Bike-sharing systems are recently becoming ubiquitous in most cities, as an environmentally friendly alternative to other means of transportation. An optimal management of the bike-sharing service would in principle benefit from the availability of a mathematical model underlying the system. Accordingly, in this paper we propose a Markov-chain based approach to model the bike-sharing system, which we believe has a potential to develop alternative methods to implement classic control actions in a bike-sharing system (e.g., in terms of implementing alternative relocation strategies or planning advertising campaigns). The proposed methodology is validated on real data from the bike-sharing system in Boston, USA, and a first application of the proposed model is preliminarily illustrated in the paper.