Characterizing Regional Importance in Cities with Human Mobility Motifs in Metro Networks

TL;DR

The paper addresses the limitation of first-order aggregated mobility networks in cities by introducing higher-order mobility motifs as fundamental units. It proposes two motif-centric constructions: a motif-based network $G_{motif}$ that aggregates induced complete graphs from individual mobility motifs, and a motif-wise network $G_{page}$ that adopts a PageRank-inspired weight redistribution from the initial station to destinations, with weights $w$ either 1 or 1/n. Using large-scale metro smart-card data from Shanghai, Beijing, and Hangzhou and house prices around stations as ground truth, the authors show that motif-based networks outperform the classic network $G=(V,W)$ in characterizing regional importance, with the motif-wise framework delivering the best performance. This demonstrates that higher-order mobility motifs are effective units for urban mobility analysis and can improve planning and policy by better reflecting travel purpose and regional attractiveness.

Abstract

Uncovering higher-order spatiotemporal dependencies within human mobility networks offers valuable insights into the analysis of urban structures. In most existing studies, human mobility networks are typically constructed by aggregating all trips without distinguishing who takes which specific trip. Instead, we claim individual mobility motifs, higher-order structures generated by daily trips of people, as fundamental units of human mobility networks. In this paper, we propose two network construction frameworks at the level of mobility motifs in characterizing regional importance in cities. Firstly, we enhance the structural dependencies within mobility motifs and proceed to construct mobility networks based on the enhanced mobility motifs. Secondly, taking inspiration from PageRank, we speculate that people would allocate values of importance to destinations according to their trip intentions. A motif-wise network construction framework is proposed based on the established mechanism. Leveraging large-scale metro data across cities, we construct three types of human mobility networks and characterize the regional importance by node importance indicators. Our comparison results suggest that the motif-based mobility network outperforms the classic mobility network, thus highlighting the efficacy of the introduced human mobility motifs. Finally, we demonstrate that the performance in characterizing the regional importance is significantly improved by our motif-wise framework.

Figures (10)

• Figure 1: Topological structures and surrounding house prices of (a) Shanghai, (b) Beijing, and (c) Hangzhou Metro networks. Each node represents a metro station. The node color reflects the weighted average house price around a metro station, and the color bar shows the corresponding price in RMB. Newly built stations that haven't been color-filled are beyond the scope of this study. Few areas around metro stations (such as airports and industrial parks) do not have residential transaction records, which will be considered in the calculations of node importance but excluded in final ranking comparisons.
• Figure 2: Illustration of three strategies of human mobility network construction. Consider a metro system with 5 stations (A-E) and 3 passengers (1-3). (a) Table list of nine Origin-Destination (OD) pairs extracted from raw transportation card data of three passengers in one day. Records with the same card ID have been sorted by time. Different colors are used to distinguish passengers and their trips. (b) Visualization of three mobility motifs generated by OD pairs. To clarify the direction of mobility motifs, the initial station of travel paths is marked in red. (c) Topological structures and corresponding adjacency matrices of human mobility networks with three different constructing strategies from mobility motifs: ($\alpha$) classic mobility network, ($\beta$) motif-based mobility network, and ($\gamma$) motif-wise mobility network. The edge color represents the passenger who contributes to the edge.
• Figure 3: Distributions of the 15 most common daily human mobility motifs in Shanghai, Beijing, and Hangzhou. The statistics of mobility motifs on weekdays and weekends are separated.
• Figure 4: The reorganization strategy from human mobility motifs $M$ to $\mathbb{M}$.
• Figure 5: Correlation between Shanghai house price and calculated station rankings in $G$ and $\mathcal{G}$. Different rows show the ranking results by PageRank, eigenvector centrality, current flow closeness centrality, and clustering coefficient, respectively. The left two columns respectively depict the correlations of the weighted average house price against a measure of the rankings of metro stations in $G'$ and $\mathcal{G}$; 95% confidence intervals are displayed. The Pearson correlation coefficient is shown in each subgraph. The right column compares $NDCG_k$ of $G'$ and $\mathcal{G}$. In order to present a clearer comparison, the inset window shows $NDCG_k$ of $\mathcal{G}$ minus $NDCG_k$ of $G'$. The orange dotted line is the datum line with $y=0$.