This a theoretical question, so I apologize if this is not the appropriate place to ask it.
Is it possible to apply Spatial LAG regression towards a categorical variable? In my experiment there are all the premises to apply spatial lag - my hypothesis is exactly that all areas influence one another, and all areas have a certain number of features on which to perform the regression - but the dependent variable is categorical rather than continuous. Is it possible to apply spatial lag in this case? Could you provide any reference? And, if LAG is not applicable, are you able to suggest a valid method for a similar check?
My goal is to corroborate the results of the calculation of Global Moran's I (that I computed for each feature), that already show a non-random spatial distribution of the features. I'd like to go a bit deeper in my analysis: in the case of a continuous target variable, I would have provided a spatial lag regression.
I am mainly working with geopandas and pysal, so any method that has an implementation in pysal is very welcome.