The MaxentVariableSelection package is useful to identify the most important set of uncorrelated environmental variables on a MaxEnt Model and also helps tune MaxEnt settings. The aim of this RStudio gadget is to simplify the tasks to select input directories and files, and MaxEnt parameters. It also helps with the creation of input files in SWD format.
At the beginning of the script are defined the default directories, files, and parameters. It is recommended that you modify them according to your OS and experiments. Please note that if you are using a computer with Windows OS, you should put the paths with the forward slash "/" or the double backwards slash "\\".
Just donwload the file gimvs.R from: https://github.com/geoabi/gimvs
Also install the following packages:
install.packages("shiny")
install.packages("miniUI")
install.packages("raster")
install.packages("MaxentVariableSelection")You can also install MaxentVariableSelection using:
devtools::install_github("alj1983/MaxentVariableSelection")You can get the last version of MaxEnt and documentation here. This gadget was tested using the version 3.3.3k that you can get here.
Open RStudio and set your working directory to the location where you downloaded the gimvs.R file. Later use the command source to execute the gadget. For example, if you donwloaded the file to C:/myScripts use:
source("C:/myScripts/gimvs.R")Alternatively you can just open the file on the file editor of RStudio and click on Source.
This gadget is an implementation of the instructions provided with the documentation of MaxentVariableSelection, you could test this gadget by following the tutorial in MaxentVariableSelection.pdf that is located in the doc directory of the package installation. You can also get it here. Please note that running the variable selection could take a long period of time depending to the input data and your computer configuration.
If you have problems running the script or you have suggestions, please open an issue here at GitHub. Also if you are interested in collaborate, please make a pull request.
