Space Time Cube analytics in QGIS and Python for hot spot detection

Çalışkan M., ANBAROĞLU B.

SoftwareX, vol.24, 2023 (SCI-Expanded) identifier

  • Publication Type: Article / Article
  • Volume: 24
  • Publication Date: 2023
  • Doi Number: 10.1016/j.softx.2023.101498
  • Journal Name: SoftwareX
  • Journal Indexes: Science Citation Index Expanded (SCI-EXPANDED), Scopus, Compendex, INSPEC, Directory of Open Access Journals
  • Keywords: Computational reproducibility, SpaceTimeCube, Space–time cluster, Spatiotemporal analysis
  • Hacettepe University Affiliated: Yes


The widespread use of Global Navigation Satellite System (GNSS) receivers for monitoring people, vehicles, and animals has generated large amounts of space–time point data. One of the important analyses of such data is hot spot detection. This could be realised by first aggregating the data into a Space Time Cube (STC), and then applying statistics like Getis–Ord Gi∗ or local Moran's I on the cells of the STC. Existing open-source software to realise this either focuses on the spatial aspect of the phenomenon or does not provide a Graphical User Interface (GUI). This paper proposes, i) a QGIS plugin, and ii) a Python package to determine the localised clusters in space and time by aggregating data into an STC. While the former provides a GUI that is easy to use, the latter is suitable for sensitivity analysis. The experiments on the openly available New York City taxi data demonstrate that the detected hot spots change depending on whether the number of passengers is used as the weight for each data point, and the statistic (i.e. Getis–Ord Gi∗ or local Moran's I). The main advantages of the developed software are two-fold. First, they contribute to free and open-source software for geospatial (FOSS4G). Second, users of varying expertise can utilise them on a potpourri of use cases ranging from transportation to criminology.