Space Time Getis Ord

Question

Hi

I posses point incident data, where the value for each point represents 1 and only 1 event. The nature of this data means that the usual method of space-time getis ord outlined here fails due to there being 0 variance. Classical ways round this would be to aggregate data or to use grids but this would remove the temporal dynamics of the underlying data. Are there any suggestions of possible ways round this other than breaking down data into time periods and performing a spatial getis ord on each time period?

Cheers

PhilipGlasner1 · Answer

Hi,

Lauren Rosenshein‌ was using a tool called "Temporal Collect Events" in this video: What's New in Spatial Statistics: Space-Time Cluster Analysis | Esri Video‌ (5:30min). I have found it "somewhere" online but looking urgently for the reference.

Can anyone help?

Philip

Public_HealthObservatory · Answer

I am in this exact same boat. I have disease data, where every record represents a person with a certain disease at a specific point in time. I can generate the spatial weights matrix using the space-time window, but cannot run Gi* because my input values are all 1 (since no two people were diagnosed with the same disease on the same date). Collecting events obviously strips the temporal aspect. This is my first attempt at this type of space-time analysis so perhaps I am missing something obvious, but what is the solution to this? 
Thanks 
Tracy