532 / 2019-02-28 19:31:07

Do different mobility datasets reveal a similar city structure?

urban mobility,taxi,mobile phone,public transit,city structure

全体主题 > Big data in Urban Planning and Management

Delineating travel patterns and understanding the city structure beneath them have long been a core research interest in urban studies. In recent years, the availability of smart card records, GPS trajectories and mobile phone data has provided unprecedented opportunity for urban researchers to investigate travel patterns and spatial interaction patterns within the city at finer spatiotemporal resolutions. Despite the abundant research efforts, most studies insofar have used a single mobility dataset and thus only showed a biased story of urban mobility. How different mobility datasets may affect our understandings of city structure has not been fully understood. In this study we attempt to investigate whether different mobility datasets might reveal a similar city structure or not using a case study in Wuhu, China. Smartcard, taxi trajectories and mobile phone data were used to delineate travel patterns and spatial interaction communities respectively. We first pre-processed the data and extracted origin-destination patterns from the three datasets. We then built spatially embedded networks for the three datasets to model the intra-city interaction patterns. Networking embedding methods and community detection methods were both used to extract spatial interaction communities and to compare how the two methods may affect results differently. Several network measures were then used to investigate the properties of sub-regions. The results suggest that first, travel patterns revealed by mobile phone and public transit usage are quite different on weekdays and weekends, while those derived from taxi trips are more similar, comparatively. Second, the travel patterns extracted from taxi and public transit usage show some correlations, which are quite different from those derived from mobile phone data. Yet further investigations on network measures suggest that public transit usage patterns have a higher level of spatial heterogeneity. By incorporating the POI dataset which provides insights on land use patterns, we found that taxi and public transit play different roles in serving certain places/population groups in the city, while mobile phone dataset may show a more comprehensive pattern of urban mobility. Our study contributes to the studies of investigating city structure using emerging data sources by demonstrating how different data may reveal different stories of intra-city interaction patterns and by addressing the importance of comparatives studies in urban analytics.