Scientific Reports (Sep 2024)

A framework for spatial-temporal cluster evolution representation and analysis based on graphs

  • Ivens Portugal,
  • Paulo Alencar,
  • Donald Cowan

DOI
https://doi.org/10.1038/s41598-024-72504-x
Journal volume & issue
Vol. 14, no. 1
pp. 1 – 21

Abstract

Read online

Abstract Analysis on spatial-temporal data has several benefits that range from an improved traffic network in a city to increased earnings for drivers and ridesharing companies. A common analysis technique is clustering. However, most clustering techniques consider a constrained representation of clusters that is limited to a single timestamp. Evolving clusters exist through several timestamps and interact with other spatial-temporal objects or clusters, having cluster relationships. When many evolving clusters exist for analysis, a graph can be used to represent cluster evolution. In this article, we propose a framework for graph-based cluster evolution representation and analysis. The framework represents cluster structure and relationships as well as provides a graph representation of cluster evolution for analysis. Evaluation is done in three case studies with spatial-temporal data about taxis that can identify important phenomena or trends in movements in a city for traffic improvement.