figshare
Browse
2_4_Yuan_Luo.pdf (936.59 kB)

Temporal analysis of multilateral spatial interactions. GeoComputation 2019

Download (936.59 kB)
Version 2 2019-12-01, 23:27
Version 1 2019-09-18, 03:00
conference contribution
posted on 2019-12-01, 23:27 authored by May YuanMay Yuan, Wei Luo
Conventional gravity-model estimates of spatial interactions assume that spatial interactions or flows from all locations to all other locations are independent of other flows and, therefore, overlook the latent multilateral influences in play. This research developed a novel application of Self-Organizing Map (SOM) to retrospectively examine historical spatial interactions among multiple locations for new insights into the spatial and temporal dependence of flows across locations. We built a SOM with units of spatial interaction patterns and traced changes in spatial interactions patterns for each location over time. By tracing the changes, we created new trajectories of spatial interactions on the SOM to contextualize spatial interactions at individual locations and among all locations over time. We used international trade data among 207 countries from 1900 to 2014 to demonstrate the proposed data-driven approach for retrospective analysis of spatial interactions. We created a SOM of international trade patterns, mapped each country’s trading trajectory, and compared the trajectories among all 207 countries. We showed that the SOM application could answer questions about multilateral spatial interactions over time. Our findings extended earlier network analyses of the global system with an integrated space-time view of spatial interactions. The SOM approach is adaptable to other domains of spatial interactions (e.g., urban transportation, immigrations) to characterize spatial interactions among locations change over time individually or contextually among all locations.

History

Publisher

University of Auckland

Usage metrics

    GeoComputation 2019

    Licence

    Exports

    RefWorks
    BibTeX
    Ref. manager
    Endnote
    DataCite
    NLM
    DC