I've tried out the Hot Spot Map tutorial for ArcMap, which is working just fine, and I'm now starting to make some modifications so that it fits my data sets. Instead of aggregating the points as in the tutorial I added a column for the weighting.

The result, using the Hot Spot Analysis (Getis-Ord Gi*) (Spatial Statistics) tool, can be seen in this picture: enter image description here

What I'm trying to do now is to repoduce these results via MATLAB, just to get an idea of how the algorithm, which is described here, works. If I'm getting it right, the GI-score is calculated from the distance between one feature and all the others and their weighting.

Using these formulas I came to the following result: enter image description here

As you can see, it looks like all the hot spots and cold spots are located in stripes.

Does ArcMap use a different algorithm from what is stated in the documentation or am I missing out on something else?

FYI, I used the default settings for the Hot Spot Analysis tool.

  • The documentation you have linked to is from ArcGIS Pro and is not from ArcMap. In any event I'm confused about whether your question is aimed at determining the algorithm used by ArcGIS for Desktop (in which case I think you should ask Esri) or understanding your MATLAB divergence from them. I think you should more clearly focus your question on whichever product you are focussing on. Your question displays results from both products but does not include the algorithm used to produce them. – PolyGeo Jul 29 '16 at 5:25