Smart and Connected Community Detection using Travelers Point of Interest

With the wide adoption of GPS-enabled smartphones and increasing availability of high-resolution GPS traces from individuals in metropolitan areas, there is an opportunity to infer sophisticated travel activity patterns and trends. This opportunity also triggers the rethinking of the traditional community detection methods to employ innovative approaches to define and form new solutions. Considering smart and connected communities, planners and advocates have begun to seek how to detect social structures that are neither individualistic nor holistic, but fundamentally relational, which can respond to their needs effectively. This study unveils smart and connected community structures using community detection methods based on the user-activity network collected by Mteropia incentive-based application. The authors test Markov graph clustering, Fast Greedy, Infomap, Leading Eigenvector, Multi-level, Edge Betweeness, Spin Glass, and Walk-Trap methods on the daily trip information of 438 anonymous users located in Tucson, AZ over 300 days. The authors further assess the community detection methods using ground truth carpooling data comprising 78,433 successful verified trips with distinct 1,129 passengers and 1,179 drivers. The results indicate Markov graph clustering outperforms other methods in revealing community structures. The outcome of this study can be used in applications that require a shift toward more demanding citizenry at a time when the collective institutions (market, state, civic features) would appear to have become less able to satisfy users demands, such as shaping carpooling networks when public transit fall short meeting the demand.

  • Supplemental Notes:
    • This paper was sponsored by TRB committee ABJ70 Standing Committee on Artificial Intelligence and Advanced Computing Applications.
  • Corporate Authors:

    Transportation Research Board

    ,    
  • Authors:
    • Arian, Ali
    • Ermagun, Alireza
    • Chiu, Yi-Chang
  • Conference:
  • Date: 2019

Language

  • English

Media Info

  • Media Type: Digital/other
  • Features: Figures; Maps; References; Tables;
  • Pagination: 18p

Subject/Index Terms

Filing Info

  • Accession Number: 01697453
  • Record Type: Publication
  • Report/Paper Numbers: 19-03098
  • Files: TRIS, TRB, ATRI
  • Created Date: Dec 7 2018 9:28AM