Jun.-Prof. René Westerholt recently presented a paper at the GISRUK (GIS Research UK) conference in Glasgow, UK. The paper deals with one of the most popular methods for detecting geographic hotspots: the Getis-Ord statistic. This statistic is a measure for revealing geographic structures in the tails of an attribute value distribution, that is, for identifying clusters of particularly high or low attribute values. The original 1995 paper presenting this measure contains an interpretation of how the variance of the spatial weights used is incorporated into the local measures. However, it can be shown that the corresponding mathematical expression is flawed. The short paper presented at the conference corrects this expression and discusses the implications of the revised term for the link between the Getis-Ord statistic and the local variances of the spatial weights.
Full reference to the paper:
Westerholt, R. (2023): A revised understanding of the influence of the variance of spatial weights on the Getis-Ord statistic. 31st Annual GIS Research UK Conference (GISRUK), Glasgow, UK. DOI: 10.5281/zenodo.7825279.