Using the Hotspot Analysis tool for Getis-Ord Gi*
Included in this tutorial:
Accessing the Hotspot Analysis tool for the Getis-Ord Gi* test for local clustering
Reviewing the tool’s parameters and options
An example with polygon features, using different weighting methods and reviewing the results
Software version in examples: ArcGIS Pro 3.4.2
Tutorial Data: The tutorial includes demonstration with sample data available here.
Credits: L. Meisterlin with Varisa Tanti and Nikolas Michael (2022); Updated by L. Meisterlin (2025)
This tutorial demonstrates using the Hot Spot Analysis (Getis-Ord Gi*) tool in ArcGIS Pro.
Accessing the Getis-Ord Gi* Tool
The path to open up the Getis-Ord Gi* tool is Geoprocessing Tools > Spatial Analysis Tools > Hot Spot Analysis (Getis-Ord Gi*).
Access the Getis-Ord Gi* tool by clicking the Geoprocessing toolbox in the main ribbon, and clicking through Spatial Analysis Tools > Mapping Clusters > Hot Spot Analysis (Getis-Ord Gi*). You can also search for “Getis-Ord Gi*” or “Hotspot” in the search bar.
Getis-Ord Gi* Parameters & Options
The inputs for running the Getis-Ord Gi* geoprocessing tool
In the Getis-Ord Gi* dialogue box, you will see these fields: Input Feature Class, Input Field, Output Feature Class, Conceptualization of Spatial Relationships, Distance Band or Threshold Distance, Self-Potential Field, and a checkbox for Apply False Discovery Rate (FDR) Correction.
A brief description of the input parameters is provided below.
The Standard Inputs
Input Feature Class: Specify the layer as your input feature data. The options that appear in the drop-down menu will be the layers currently in your map’s Contents. You can also choose to browse for an input file (by clicking the yellow folder icon).
Input Field: The numeric field to be evaluated from the input feature class’s attribute table.
Output Feature Class: Name and indicate the location of the output feature class.
Conceptualization of Spatial Relationships: This option defines the approach to spatial weighting between features in the input feature class. The drop-down menu offers different approaches.
Inputs dependent upon the Conceptualization of Spatial Relationships
Depending on the option selected for the Conceptualization of Spatial Relationships (spatial weights), different options will appear.
for distance-based spatial weighting methods
Distance Method: Choose a straight-line (Euclidean) or cardinal-right-angle (Manhattan) method of measuring distance.
Distance Band or Threshold Distance: Specify a cutoff (or “threshold”) distance in the units of the input feature class’s CRS.
for weighting by K nearest neighbors
Number of Nearest Neighbors: Specify the number of nearest features to include in the clustering calculation.
to use a different (custom) weighting approach
If you choose “Get spatial weights from file” under the Conceptualization of Spatial Relationships option, you must supply a matrix (table) file under Weights Matrix Table option.
Optional Inputs
Self Potential Field: The distance or weight between a feature and itself.
Apply False Discovery Rate (FDR) Correction: Specifies whether statistical significance will be assessed with or without False Discovery Rate (FDR). When FDR is checked, the critical p-values determining confidence levels is reduced to account for multiple testing and spatial dependence.
Example: Polygon Features
This example will use the Standard Data Package from Tutorial Data to calculate Getis-Ord GI* analysis. The standard data package for the tutorials includes randomly generated points with randomly generated values in the attribute table. As a result, this tutorial demonstrates using this tool and examining its results, but we do not expect to find strong clustering patterns.
The input parameters include
Input feature class: the blocks features, and
Input field: the Int_lrg attribute values.
For reference, the feature class and its attribute values are symbolized below.
example inputs
Example 1: Contiguity with Edges and Corners
For the first example, we run the tool comparing contiguous features, using both edge- and corner-contiguity.
Conceptualization of Spatial Relationships: Contiguity edges corners.
Examining the Results
The default symbology of results in the map view represents the results in a color ramp from blue (cold spots—clusters of low values) to red (hot spots—clusters of high values), ranked by their p-values (confidence level).
edges-and-corner contiguity results, mapped
By default, the attribute table of the resulting feature class includes
the input value per feature used as the input field (here, it is the Int_lrg field),
the Gi* value per feature (GiZScore),
a p-value (GiPValue),
the number of other features considered within the analysis (NNeighbors), and
a coded field that is used to group the features based on their Gi* value and p-value in to the “bins” that are symbolized in the default symbology, above (Gi_Bin).
edges-and-corner contiguity results, in the attribute table
Example 2: K-Nearest Neighbors
Next, we run the tool a second time with a different Conceptualization of Spatial Relationships (spatial weight) option to compare the results. The input feature class and input field are unchanged from the previous example.
Conceptualization of Spatial Relationships: K nearest neighbors
Number of Neighbors: 12
Examining the Results
Once again, the default symbology of results is represented below, in a color ramp from blue (cold spots–clusters of low values) to red (hot spots–clusters of high values), ranked by their p-values (confidence level). Notice the difference in results from Example 1, despite using the same input geometry and values.
12-nearest neighbor results, mapped
The attribute table includes similar information about the input value and the results. Here, each feature’s results were the obtained by comparing 13 different values (the input feature’s value + its 12 nearest neighbors).
12-nearest neighbor results, in the attribute table