Implementing GLMMs in ArcGIS

Question

I am looking to create a predictive model in ArcGIS. I have done this previously with glm with poisson distribution and found it relatively straightforward in raster calculator e.g.

exp(intercept + (xvariable) + (yvariable))

However, it is likely that i shall need to use a more complex approach to account for both local and landscape variables and will need to use a Generalised Linear Mixed Model.

Does anyone have experience of this? Is it possible to do in ArcGIS(raster calculator), how would the equation likely differ to that above?

Answer 1

You might be able to create a python geoprocessing script for ArcGIS, and load in some python libraries to do the stats for you. There are a lot of stats libraries out there, a good place to start would be:

http://statsmodels.sourceforge.net/devel/glm.html

I also remember reading about how you can create R scripts within python as well. A user would need to have both ArcGIS and R installed and licensed on their system to use it. I haven't done it yet, but am seriously thinking about trying to use R with ArcGIS in the near future, so I'd be curious how it goes if you go that route!