The hardware and bandwidth for this mirror is donated by METANET, the Webhosting and Full Service-Cloud Provider.
If you wish to report a bug, or if you are interested in having us mirror your free-software or open-source project, please feel free to contact us at mirror[@]metanet.ch.
ENMeval
is an R package that performs automated tuning and evaluations of
ecological niche models / species distribution models. These models make
predictions of species’ niche relationships and potential geographic
distributions based on presence data, environmental predictor variables,
and a sample of available environmental conditions (i.e., background
data).
“Model tuning” is commonly used for machine-learning models. It means building candidate models with a range of complexity settings, evaluating the accuracy of each one (here with cross-validation), then selecting optimal settings for your data based on those of the best-performing model. This exercise is important because it is difficult to predict in advance how complex your model needs to be to make accurate and ecologically realistic predictions for your species. Too much model complexity leads to overfitting, where your model fits your data very well but it cannot predict new data accurately. Model tuning helps maximize predictive ability while avoiding model overfitting.
The ENMeval package features a single function that
performs model tuning based on user specifications, including methods
for partitioning data for cross-validation (random, leave-one-out,
spatial, custom), and evaluates models using predefined performance
metrics (AUC, Continuous Boyce Index, omission rates) with the option to
insert others. The package includes functionality for three models: maxent.jar
(Java implementation of Maxent), maxnet (R implementation
of Maxent), and BIOCLIM
(climate envelope method). Users can also specify other algorithms by
customizing an ENMdetails object
(?ENMdetails). The package also offers comprehensive
metadata output, null model evaluations, visualization tools, and an
extensive tutorial
that walks you through a full analysis workflow. Many features in
ENMeval were created in response to user requests – thank
you for your input! Version >=2.0.0 represents an extensive
restructure and expansion of previous versions, and 2.0.5 is a big move
from raster and dismo functions to those of
terra and predicts.
For a more detailed description of ENMeval, please
reference the most recent publication:
For the original package version, please reference this older publication:
ENMeval is a work in progress, changing slowly to
fix bugs when users identify them. If you find a bug, please raise an
Issue in this Github repo and I will resolve it as soon as I can. The
CRAN version may lag behind the Github one, so please try the
development version here first if you are having any issues. Install
with:
devtools::install_github("jamiemkass/ENMeval")
The vignette is not included in the CRAN version of the package due to file size constraints, but is available on the package’s Github Pages website.
Please make sure to use the most recent version of maxent.jar, as bug fixes are occasionally made.
Note that as of version 0.3.0, the default implementation uses the ‘maxnet’ R package. The output from this differs from that of the original Java program and so some features are not compatible (e.g., variable importance, html output).
These binaries (installable software) and packages are in development.
They may not be fully stable and should be used with caution. We make no claims about them.