Exploring the spatiotemporal structure of dynamic urban space using metro smart card records

The wide application of pervasive computing technology has allowed for the emergence of big data on spatial behavior and therefore provides an opportunity to explore dynamic urban space. In this paper, an eigendecomposition method is proposed to capture the common patterns of passengers' variation over time among all metro stations as well as to explore the spatial heterogeneity of the dynamic space around the metro stations based on the common patterns with low dimensional structures. Using Shenzhen as a case study, four datasets for check-in/check-out and weekday/weekend are decomposed to obtain the principal components (PCs) and eigenvectors. The first several PCs are the most common patterns of passengers' variation over time among all metro stations, while the corresponding elements in the eigenvectors, referred to as EigenStation in this paper, can describe the characteristics of the metro station. The decomposition result is evaluated at both the aggregation and individual station levels, and the result demonstrates that the first two elements of the EigenStation can approximate the original dataset. The EigenStation vector angle, i.e., ω, is used to represent the structure of the EigenStation, and its value is highly related to the land use structure around the metro stations. The proposed method can provide deep insight into static and dynamic urban spaces, which can help improve urban planning around metro stations.