How are 90%, 95%, and 99% confidence levels used when assigning Emerging Hot Spot Analysis categories in ArcGIS Pro?

Question

Hi, I'm using Emerging Hot Spot Analysis (EHSA) in ArcGIS Pro 3.6 for a research project.

From my understanding, the workflow is:

A Getis-Ord Gi* statistic is computed for every space-time bin. ArcGIS applies False Discovery Rate (FDR) correction and labels each bin as: 99% confidence hotspot 95% confidence hotspot 90% confidence hotspot not significant (<90%) EHSA then assigns final categories such as New, Consecutive, Persistent, Sporadic, Historical Hot Spot, etc.

My question is specifically about step 3.

How are the 90%, 95%, and 99% confidence levels incorporated when populating the final EHSA categories?

For example, suppose a location has these sequences over 12 time steps:

A: 0, 0, 0, 0, 0, 0, 0, 0, 0, H(90), H(90), H(90)

B: 0, 0, 0, 0, 0, 0, 0, 0, 0, H(99), H(99), H(99)

where:

H(90) = hotspot at 90% confidence H(99) = hotspot at 99% confidence 0 = not significant

My question is:

If both sequences are classified as Consecutive Hot Spots, are they treated identically by EHSA? Does EHSA first convert all statistically significant hotspots (≥90%) into a simple hotspot/non-hotspot (binary) state before applying the temporal logic? Or do stronger confidence levels (95% and 99%) receive additional weight when assigning the final categories?

Any references, ESRI technical notes, or implementation details would be greatly appreciated.

Robert_LeClair · Answer

EHSA does not apply extra “weight” to 99% vs 90% when it assigns the pattern categories (New/Consecutive/Persistent/Sporadic/etc.). For the pattern logic, each time step is effectively treated as hot / cold / not significant based on the Gi_Bin significance classification (after FDR, if you enabled it). The strength (90 vs 95 vs 99) affects whether a bin is considered significant, but once it is significant, the temporal pattern rules don’t further differentiate 90% from 99% for category assignment.

So in your examples:

* A: …, H(90), H(90), H(90)
* B: …, H(99), H(99), H(99)

If both meet the definition of Consecutive Hot Spot (final time step is a hot spot and there’s an uninterrupted run of hot spots at the end, with the run not long enough to be “persistent”), then EHSA will categorize them the same. The 99% sequence doesn’t get special priority for being “more confident,” aside from being less likely to flip to not significant.

The “fuzzy weighting between 90/95/99” concept exists in a different tool: Hot Spot Analysis Comparison, which explicitly provides options like Combine 90/95/99 significant, Exact matching, etc., and documents the similarity weighting approach. EHSA’s pattern categories are not based on those similarity weights.

DanielFox1 · Answer

Hi @Erfankt

Here are links to documentation which you might find useful.

How Emerging Hot Spot Analysis works | ArcGIS Pro documentation

Hot Spot Analysis (Getis-Ord Gi*) (Spatial Statistics Tools) | ArcGIS Pro documentation