Recently, free floating bike sharing (FFBS) becomes prevalent in China because of its properties of convenience, sustainability and energy savings. Better than traditional bike sharing, FFBS can be almost obtained and parked at any available places, not limited by the location of docking station. However, traffic chaos caused by unbalanced allocation and disorderly parking also emerges at the same time. Accurate FFBS traffic patterns exploration is the key to active traffic management. To achieve it, a large-scale travel records and the online points of interests (POIs) in Shanghai are collected in this study. Using these data, 3,325 FFBS gathering areas are discovered and six typical categories of land use are extracted by clustering analysis. Then Latent Dirichlet Allocation (LDA) is conducted to discover the latent FFBS travel patterns and typical travel patterns are discussed in temporal and spatial characteristics. The results show that huge difference in travel patterns exists between weekdays and weekends, and arrival times in a day are not evenly distributed. The research results are beneficial for reasonable resource allocation and helpful to realize the accurate traffic management for FFBS.

CICTP