project: Project a fitted Maxent model
In johnbaums/rmaxent: Tools for working with Maxent in R

Description Usage Arguments Details Value Warning References See Also Examples

Project a fitted Maxent model by predicting to new environmental data.

1	project(lambdas, newdata, return_lfx = FALSE, mask, quiet = FALSE)

`lambdas`	Either (1) a `MaxEnt` fitted model object (fitted with the `maxent` function in the `dismo` package), (2) a file path to a Maxent .lambdas file, or (3) a `lambdas` object returned by `parse_lambdas()`.
`newdata`	A `RasterStack`, `RasterBrick`, `list`, `data.frame`, `data.table`, or `matrix` that has layers/elements/columns whose names correspond to the names of predictors used to fit the model. These layers/elements/columns must all have the same length.
`return_lfx`	Logical. Should `Raster` layers be returned giving lambdafeature values for each feature with a non-zero lambda? Currently ignored if `newdata` is not a `Raster` object.
`mask`	(Optional; requires that `newdata` is a `Raster` object.) A `Raster` object with `NA` values in cells for which the model should not* be projected. These cells will be assigned `NA` in the returned output.
`quiet`	Logical. Should projection progress be reported?

project uses feature weights described in a .lambas file or MaxEnt object to predict a Maxent model to environmental data. This function performs the projection entirely in R, without the need for the Maxent Java software. For tested datasets, it performs the projection in roughly one third of the time taken for the same projection by maxent.jar.

If newdata is a RasterStack or RasterBrick, a list with three elements:

prediction_raw: a Raster layer giving the raw Maxent prediction;
prediction_logistic: a Raster layer giving the logistic Maxent prediction; and
prediction_cloglog: a Raster layer giving the cloglog Maxent prediction. If newdata is not a RasterStack or RasterBrick, the raster layers will be replaced with data.tables in the returned list.

Additionally, if newdata is a RasterStack or RasterBrick and return_lfx is TRUE, the returned list will include prediction_lfx (the logit scores for the linear predictor), and lfx_all (the contributions to prediction_lfx of each feature with a non-zero lambda).

This function is still in development, and no guarantee is made for the accuracy of its projections.

Wilson, P. W. (2009) Guidelines for computing MaxEnt model output values from a lambdas file.
Maxent software for species habitat modeling, version 3.3.3k help file (software freely available here).

read_mxe()

## Not run: 
# Below we use the dismo::maxent example to fit a Maxent model:
fnames <- list.files(system.file('ex', package='dismo'), '\\.grd$', 
                     full.names=TRUE )
predictors <- stack(fnames)
occurrence <- system.file('ex/bradypus.csv', package='dismo')
occ <- read.table(occurrence, header=TRUE, sep=',')[,-1]
me <- maxent(predictors, occ, factors='biome')

# ... and then predict it to the full environmental grids:
pred <- project(me, predictors)
# This is equivalent to using the predict method for MaxEnt objects:
pred2 <- predict(me, predictors, args='outputformat=logistic')
pred3 <- predict(me, predictors, args='outputformat=cloglog')

all.equal(values(pred$prediction_logistic), values(pred2))
all.equal(values(pred$prediction_cloglog), values(pred3))

## End(Not run)