maxent_IC: Calculate information criteria for Maxent models
In johnbaums/things: Mostly spatial things

Description Usage Arguments Value Warning References Examples

Calculate AIC, AICc, and BIC for Maxent models as in ENMTools.

1	maxent_IC(x, occ, lambdas)

`x`	Either a file path to a Maxent raw prediction raster, or a `Raster` or `RasterStack` containing raw Maxent predictions for which information criteria will be calculated.
`occ`	A `matrix` or `data.frame` containing the coordinates for occurrence data used to fit the model.
`lambdas`	The path to a Maxent .lambdas file, or a `MaxEnt` fitted model object.

An n x 6 matrix, where n is the number of Maxent rasters for which information criteria are to be calculated. Columns give n (the number of occurrence records used for model training), k (the number of features with non-zero weights), ll (the negative log likelihood of the model), AIC, AICc, and BIC (as calculated in ENMTools).

These information criteria should not be calculated for models that use hinge or threshold features because the number of predictors is not estimated correctly.

Warren, D. L., Glor, R. E. and Turelli, M. 2009. ENMTools: a toolbox for comparative studies of environmental niche models. Ecography 33:607-611
ENMTools

library(dismo)
fnames <- list.files(path=file.path(system.file(package='dismo'), 'ex'), 
                     patt='grd', full=TRUE )
predictors <- stack(fnames)
occ <- read.csv(file.path(system.file(package="dismo"), 'ex/bradypus.csv'))[, -1]
me <- maxent(predictors, occ, args=c('hinge=false', 'threshold=false'),
             path=tempdir())
r <- project_maxent(me, predictors, quiet=TRUE)$prediction_raw

# passing the raster object to pred.raw and the maxent object to lambdas:
maxent_IC(r, occ, me)

# passing a lambdas file path to lambdas:
maxent_IC(r, occ, file.path(tempdir(), 'species.lambdas'))

# passing a raster file path and lambdas file path to lambdas:
writeRaster(r, f <- tempfile(fileext='.tif'))
maxent_IC(f, occ, file.path(tempdir(), 'species.lambdas'))