A Data-Driven Based Dynamic Rebalancing Methodology for Bike Sharing Systems

Mobility in cities is a fundamental asset and opens several problems in decision making and the creation of new services for citizens. In the last years, transportation sharing systems have been continuously growing. Among these, bike sharing systems became commonly adopted. There exist two different categories of bike sharing systems: station-based systems and free-floating services. In this paper, we concentrate our analyses on station-based systems. Such systems require periodic rebalancing operations to guarantee good quality of service and system usability by moving bicycles from full stations to empty stations. In particular, in this paper, we propose a dynamic bicycle rebalancing methodology based on frequent pattern mining and its implementation. The extracted patterns represent frequent unbalanced situations among nearby stations. They are used to predict upcoming critical statuses and plan the most effective rebalancing operations using an entirely data-driven approach. Experiments performed on real data of the Barcelona bike sharing system show the effectiveness of the proposed approach.

Automated summary

Mobility in cities is a fundamental asset and poses challenges in decision making and service creation. Bike sharing systems, particularly station-based ones, require periodic rebalancing operations to ensure system usability. This paper proposes a dynamic bicycle rebalancing methodology based on frequent pattern mining, showing effectiveness in real data experiments.