Optimized Hot Spot Analysis - Point vs. Area problem

Discussion created by torbenschmack on Mar 5, 2014
Hi there,

I'm running the optimized hot spot tool in order to determine locations suited to host new guesthouses.
Therefore i want to exclude areas from a predefined study area (polygon) which are unsuited. The outputs from two hotspot analyses shall generate the areas which are to be excluded.

The problem:
As the standard hotspot output are points of statistically significant hot or cold spots, they are unsuited for this task since points don't really reflect an area.
Fishnet cells (which can be generated as output of hot spot analysis) would be a solution but only work if there's no analysis field specified for the point data analyzed. For one of my analyses this is the case - for the other unfortunately not.

My question:
Is it statiscally allowed to somehow create a hot spot area / surface (not a density surface!) from the hot spot points such as a buffer around the points?

Would the statistic significance also apply for the whole buffer area, if this buffer would have the distance band used for the hot spot analysis as a radius?
As both distance bands are several 100 meters the buffers would cover my whole study area and thus no positive locations would remain.

So if buffer is fine, which radius would I have to choose? ;)

Doesn't the same problem apply to hot spot analysis with fishnet cells?
How does a fishnet cell (which is an area and not a point!) as an output of a hotspot analysis maintain the statistic integrity?

Thanks for your thoughts!