Looking to fill in gaps through generalizing a population count across U.S. counties. I.e., if a county has 0 population, but is surrounded by 5 non-zero counties, then change the 0-population county to an estimate based on the neighboring non-0 counties. This is all vector polygon data, no points, no raster, etc.
I do not have the Spatial Analyst extension in ArcMap or Pro, so I have been trying the R-ArcGIS bridge extension to run a Empirical Bayes smoothing on the data. The issue is that the calculation fails when it encounters non-positive population values.
Another thought was to use the Polygon Neighbors tool which will spit out counts for all adjacent geographies, and maybe from there I could assign the average of the neighbors. Does anyone have a more precise way? Even if it's a Pythonic?