A mobility network approach to identify and anticipate large crowd gatherings. (August 2018)

• Record Type: Journal Article
• Authors: Huang, Zhiren; Wang, Pu; Zhang, Fan; Gao, Jianxi; Schich, Maximilian
• Is Part Of: Transportation research. Volume 114(2018)
• Page Start: 147
• Page End: 170
• Publication Date: 2018-08
• Subjects: Human mobility -- Large crowd gatherings -- Complex networks -- Collective behavior; Transportation -- Research -- Periodicals; Transportation -- Mathematical models -- Periodicals
• DOI: 10.1016/j.trb.2018.05.016
• Languages: English
• ISSNs: 0191-2615

Abstract

Highlights: We answer the important question of how crowds come together before they enter the crowding zone. The used 200 million individual trip records guarantee that mobility fluxes during crowding events can be traced at an unprecedented accurate and comprehensive level. We introduce the concept of anomalous mobility networks to identify essential mobility patterns feeding into large crowd gatherings. Our study contributes valuable, likely life-saving insight to anticipate and avoid potentially dangerous crowding situations. Abstract: The study of large crowd gatherings combines aspects of longer-range human mobility with site-specific pedestrian dynamics. Recently, substantial progress has been made in understanding the collective behaviors of crowds on the site-specific scale. Yet, the human mobility aspect remains vague in terms of how large crowds come together in the first place. Using high-resolution human mobility data in form of millions, potentially real-time, subway and taxi records, our approach uncovers the mobility patterns involved in large crowd gatherings. In addition, we discriminate anomalous mobility fluxes from ordinary mobility fluxes by introducing the concept of anomalous mobility networks, within which nodes are traffic zones and links are defined via the Jensen-Shannon divergence. Our approach allows for easy identification of occurrence, location and developing stages of crowd formation. Strikingly, within the anomalous mobility networks, we find high-stress crowd density to be preceded by a node in-degree k in surpassing the critical threshold kc, typically preceding the maximum crowd density by a couple of hours, enabling us to anticipate large crowd gatherings via a surprisingly simple approach based on the simple network index k in .