Interactive visual analytics for multi-city bikeshare data analysis

摘要: Bicycle sharing schemes are gaining traction as an alternative or complementary mode of urban transport. In this paper, we explore the utility of an interactive web-based visual analytics application for comparing usage patterns between different bike sharing programs. We demonstrate the potential for adjustable filters on user demographics and trip characteristics to reveal differences in ridership between cities. We also perform clique-detection and Louvain modularitybased community detection to reveal areas of high connectedness under different contexts. Our work utilizes the STDBSCAN algorithm in a novel context to cluster trips as a means of categorizing flow patterns. Finally, using publicly available data from bike share organizations, we conduct some experiments combining the data filters, algorithms and visualization. This preliminary work showcases the value of interactive visual analytics for highlighting notable differences between established bike share systems that may help frame questions for future research or policymaking.