Including offsets is another option to integrate spatial prior information in linear and additive regression models. Offsets shift the intercept of the regression fit by a certain amount. Although only one offset can be added to a regression model, it is possible to combine several spatial-explicit estimates into one offset by calculating the sum of all spatial-explicit layers.
Usage
add_offset(x, layer, add = TRUE)
# S4 method for class 'BiodiversityDistribution,SpatRaster'
add_offset(x, layer, add = TRUE)
# S4 method for class 'BiodiversityDistribution,sf'
add_offset(x, layer, add = TRUE)
Arguments
- x
distribution()
(i.e.BiodiversityDistribution
) object.- layer
A
sf
orSpatRaster
object with the range for the target feature.- add
logical
specifying whether new offset is to be added. Setting this parameter toFALSE
replaces the current offsets with the new one (Default:TRUE
).
Value
Adds an offset to a distribution
object.
Details
This function allows to set any specific offset to a regression
model. The offset has to be provided as spatial SpatRaster
object. This
function simply adds the layer to a distribution()
object.
Note that any transformation of the offset (such as log
) has do be done externally!
If the layer is range and requires additional formatting, consider using the
function add_offset_range()
which has additional functionalities such
such distance transformations.
Note
Since offsets only make sense for linear regressions (and not for
instance regression tree based methods such as engine_bart()
), they do not
work for all engines. Offsets specified for non-supported engines are ignored
during the estimation
References
Merow, C., Allen, J.M., Aiello-Lammens, M., Silander, J.A., 2016. Improving niche and range estimates with Maxent and point process models by integrating spatially explicit information. Glob. Ecol. Biogeogr. 25, 1022–1036. https://doi.org/10.1111/geb.12453
See also
Other offset:
add_offset_bias()
,
add_offset_elevation()
,
add_offset_range()
,
rm_offset()
Examples
if (FALSE) { # \dontrun{
x <- distribution(background) |>
add_predictors(covariates) |>
add_offset(nicheEstimate)
} # }