To path indicator Subpages of “Newsroom“ To navigation To quick access To footer with other services
To content
Department of Spatial Planning
A person types in something on a smartphone © FlyingFife​/​Shotshop.com

Presentation of a new ‘cluster plot’ at the GISRUK 2026 conference

Cluster plot for characterising heterogeneous spatial clusters © René Westerholt​/​TU Dortmund
René Westerholt presented a novel ‘cluster plot’ at the GIS Research UK conference 2026 in Birmingham.

Communicating spatial statistical results effectively is not always an easy task. For example, when looking for spatial clusters of high and low values for a particular attribute, the resulting maps of corresponding statistics typically display coloured map symbols that indicate whether a polygon, point, or line is in any way unusual. However, the clusters identified in such a way can be very different in nature despite sharing the same map symbols. For example, some identified clusters may be internally homogeneous, whilst others may be characterised by single outliers. Both scenarios would lead to different interpretations and are therefore of great significance for planning and geographical research as well as for practical applications. The contribution by René Westerholt, presented at the 34th annual GIS Research UK conference in Birmingham, introduces the cluster plot, a visualisation that combines the spatial autocorrelation, hotspot, and spatial variance statistics Local Moran’s I, Getis–Ord Gi*, and Local Spatial Heteroscedasticity. Integrating these features in a single plot allows for easier characterisation of spatial clusters, including the identification of homogeneous hotspots and coldspots, hotspots and coldspots driven by local outliers, and irregular clusters associated with high local variability. Such characterisations would otherwise have to be derived from multiple maps and in a less intuitive manner, making the cluster plot a valuable addition to the visual toolbox for exploratory spatial data analysis. The visualisation is currently being implemented by Martin Fleischmann (Charles University, Prague) and Levi Wolf (University of Bristol) for integration into the Python Spatial Analysis Library (PySAL), where it will soon be available under the name ‘local cross plot’.

Westerholt, R. (2026): A novel cluster plot for the visual characterisation of irregular spatial hotspots and coldspots. 34th Annual GIS Research UK Conference (GISRUK), Birmingham, UK. DOI: 10.5281/zenodo.19421814

Back