Detecting and visualizing cohesive activity-travel patterns: A network analysis approach

Abstract

This article presents a network analytical framework to detect individual-based activity-travel patterns (ATPs) in space and time. Compared to many existing classification methods (e.g., hot-spot detection, sequential alignment method), the network method substantiates the social meanings underlying the interconnectedness and similarities of people's activity trajectories and better integrates spatial interaction (colocation or distance-decay) and temporal connections (concurrence or sequence) of daily lives in the measure of similarity. This approach enables us to detect variant community structures, with individuals in the same community interacting relatively more than individuals belonging to different communities, by decomposing the complex trajectories into different meaningful events (e.g., activities, trips, tours, and subsequences). We also demonstrate the practicality and scientific merit of the network analysis approach in a case study of household travel behavior in Charlotte, North Carolina. Results derived from disaggregated survey data establish the effectiveness and flexibility of the network methods to detect cohesive communities of individuals and ATPs by different narratives of everyday-life events. This study also suggests that the network analysis approach has great potential to classify large datasets of other space-time trajectories and to discover policy-sensitive activity, trip, and tour patterns that help us develop policy and planning alternatives for sustainable communities and mobility.