A model framework for discovering the spatio-temporal usage patterns of public free-floating bike-sharing system

Public bike-sharing has gained much attention with the tide of sharing economy. Empowered by modern technologies (e.g., GPS devices and smartphone-based APPs), a new generation of free-floating bike-sharing systems has recently become popular. Usage data generated by such systems produce rich information. This study presents a model framework to explore the spatio-temporal usage patterns of free-floating shared bikes using the usage data. The framework includes modules of probability fitting, Random Forest, a cluster-based time-domain analysis, and a visualization toolset. A case study is discussed based on the usage data from Mobike, one of the largest operating bike-sharing systems in Shanghai (China). The daily usage dynamics is modeled using log-normal distributions. Random Forest is adopted to explore the impact of factors on the usage frequency in different districts. It is found that residential area, park & green area, and population size are the top three factors influencing the frequency. Particularly, usage near metro stations is delved using the hierarchical clustering method, resulting in three typical usage modes. Visualization analysis is demonstrated to understand the time-varying flow patterns and the spatial distribution of shared bikes. This study improves our understanding of the usage patterns of this emerging transport mode and provides insights for the promotion and dynamic deployment of the bike-sharing system in urban areas.