Skip to content

Spatial autocorrelation in Maxent models

2 messages · Diego Guevara, Bede-Fazekas Ákos

#
Hello everyone

I am trying to find a way to test for spatial autocorrelation in Maxent
models. From my understanding, this should be tested on the residuals of
the model. So far I have only been able to test for Moran I test on the
prediction resulting raster using the package ELSA. However, I think I am
just testing Moran on the prediction values this way. Would that be enough
or ok? I will appreciate if someone can give me some insight.

Best wishes
Diego
#
Dear Diego,

Yes, you calculated the spatial autocorrelation of the predicted values 
in this way, which is not too informative and intresting. 
Autocorrelation of the residuals is what you need.
If you subtract the predictions from the observations (using the simple 
raster algebra or raster::calc()), you can calculate Moran's I (or any 
other spatial autocorrelation measure) of the residuals.

HTH,
?kos Bede-Fazekas

2021.01.15. 12:21 keltez?ssel, Diego Guevara ?rta: