Spatiotemporal Mining of BSS Data for Characterising Seasonal Urban Mobility Dynamics

Abstract

Digital traces of individual mobility can be revealed from the origin-destination sensing systems of BSS (Bicycle Sharing System). This record enables wide analysis of human mobility traits in urban area including pattern, trend, and anomalies. This study investigates and compares trip history of BSS open data from two cities, London and New York, along a year period with respect to annual weather data as explanatory factors. This aims to get insights about seasonal urban mobility dynamics both temporally and spatially. Results show that, for both cities, there are differences as well as similarities of temporal correlation level between riding behaviour of BSS users and hour of the day, day of the week, season, and local weather. Practically, the most correlated factor can be further considered and used as predictive features. Meanwhile, the proposed spatial analysis shows the positive bikes imbalance occurs in the morning, mostly at inner stations because of inward flow, and vice versa. This spatial extent can be used for redistribution purpose, specifically in order to provide enough resources at the highly visited stations before peak time occurs.