heat map based on values instead of dot density

[RobertKossick]

I have a shapefile that has thousands of point-features spread across the country (census subdivisions) however these points are not evenly distributed (concentrated in cities). Each point has values that are percentages, based on population. Like the number of GIS users per capita. I am trying to make a heat map to show the highest points, however my cities always make the hotspots, even if my highest percentage is located out in the country.

I can't include all the features because there are too many points. Is there a way to make a heat map that ignores the city hot-spots and will show the highest percentage points as hottest? even if they are all on their own in the middle of nowhere?

[Anonymous User]

Hi Robert,

Hot Spot analysis is always looking for clusters, so the densest points will almost always show up in the analysis as hotspots:

To be a statistically significant hot spot, a feature will have a high value and be surrounded by other features with high values as well (Find Hot Spots—ArcGIS Online Help | ArcGIS ). In this case, if we only have one high percentage by itself, it's not necessarily a hotspot. Neither is a random distribution of values that are clustered spatially. 

There's a lot more information about how the Find Hot Spots tool works in the "How Optimized Hot Spot Analysis Works" documentation for ArcMap (this is the same process the Find Hot Spots tool uses):

How Optimized Hot Spot Analysis Works—Help | ArcGIS Desktop 

In particular, it discusses locational outliers, which may explain some of the behavior you're seeing (How Optimized Hot Spot Analysis Works—Help | ArcGIS Desktop - Initial Data Assessment).

I'm not sure what a good alternative would be, but maybe others out here on GeoNet have ideas! And if I think of any, I'll definitely add to the post!

Best,

-Lauren

[NickDierks1]

It sounds like rather than a hotspot visualization, you'd benefit more by symbolizing your data based on counts and amounts (size or color). Since you said a major obstacle is the high number of points, what if you filtered out the bottom X%, so you showed only higher values? It's not ideal, because your map will no longer show where low spots are, but your 'hotspots' will be clearly identified.

Or maybe you could spatial-join your data to counties or something, with each county showing the average values of your chosen attributes?

[DavidBrown5]

Hi Lauren,

Did you ever determine a method to do what you asked?  I'm looking for an answer as well.  

-David