Car hailing is undergoing rapid global development, thereby providing new opportunities and challenges to operators and transport engineers due to uneven or irregular demand in certain areas. To date, only a limited number of studies have analyzed regional mobility patterns or anomaly detection. This study therefore proposes a methodology for recognizing regional mobility patterns using car-hailing order datasets and point of interest datasets. More specifically, we detect regional mobility patterns by incorporating regional intrinsic properties to a hierarchical mixture model termed latent Dirichlet allocation (LDA). This model can simulate the process of generating car-hailing order data and yield regional mobility patterns from spatial, temporal, and spatiotemporal perspectives. Moreover, by combining the trained results with future mobility records, we can measure similarities between areas and detect anomalous areas by calculating the perplexity. We also implement our workflow on a real-word car-hailing order dataset and reveal that it is possible to identify areas with similar or anomaly mobility patterns. This research will contribute to the design of regional transportation policies and customized bus services.


    Zugriff

    Download


    Exportieren, teilen und zitieren



    Titel :

    Understanding Regional Mobility Patterns Using Car-Hailing Order Data and Points of Interest Data


    Beteiligte:
    Zheng Zhang (Autor:in) / Yanyan Chen (Autor:in) / Jie Xiong (Autor:in) / Tianwen Liang (Autor:in)


    Erscheinungsdatum :

    2020




    Medientyp :

    Aufsatz (Zeitschrift)


    Format :

    Elektronische Ressource


    Sprache :

    Unbekannt