Hotspot Spatial Analysis Overview

A hotspot analysis is used to identify clusters of locations that have disproportionately high or low values. For example, a hotspot analysis could be used to identify if a focal county and its neighboring counties have a disproportionately high prevalence of diabetes compared to all counties in the state. A hotspot analysis is based on Getis-Ord statistics. Specifically, the USDSS uses the \(G_i^*\) statistic, as it is more commonly used in spatial pattern analysis (Getis & Ord, 1992). Generally, lower values of \(G_i^*\) may indicate clusters of low values, while higher values of \(G_i^*\) may indicate clusters of high values.

A \(G_i^*\) statistic is calculated for each areal unit (e.g., a county) under consideration and may be standardized as a \(z\)-score to aid in interpretation. For positive z-scores, the larger the value, the more likely it is to indicate a hotspot; for negative z-scores, the smaller the value, the more likely it is to indicate a coldspot. Given certain assumptions, the z-score form of the \(G_i^*\) statistic can be used to assess statistical significance or it can be used more generally to visualize spatial patterns.

For more information, please visit the FAQ.

Reference:

Getis A, Ord JK. The Analysis of Spatial Association by Use of Distance Statistics. Geographical Analysis. 1992;24(3):189-206. doi:10.1111/j.1538-4632.1992.tb00261.x