trainMaxentModels: Train maxent models with the provided occurence data,...

In jscavetta95/maxentools: Species Distribution Modeling with Maxent

Description

Train models with varying factor combinations and beta regularization values. Each factor combination provided will be separately trained the desired amount of beta regularization values. Beta regularization values are found using the LIPO search technique, which will generally tend towards the best hyperparameter value for the test case. Many metrics are calculated and provided, however, currently the symmetric extremal dependence index (SEDI) metric is used to determine the best model. Total number of models trained = factors * regularization iterations.

Usage

trainMaxentModels(occurrence_data, background_data, predictors, factors,
  regularization_iterations, regularization_range, shape, output_folder)

Arguments

occurrence_data	An occurrence dataframe containing cooridnates and environmental data at each coordinate.
background_data	Dataframe of selected background coordinates and extracted environmental predictor data.
predictors	A RasterStack of RasterLayer objects representing the environmental predictors.
factors	The combination of factors to use. Should be structured like "LQHPT". L = linear, Q = quadratic, H = hinge, P = product, T = threshold.
regularization_iterations	The number of regularization values to try for each factor value. Essentially the number of models to run for each factor.
regularization_range	A 2 element vector containing the start and end of the regularization values to consider in the search.
shape	A shapefile to overlay on produced predictive maps.
output_folder	The path where data and plots will be saved.

Value

A list containing the best model, the occurrence data used, the arguments (hyperparameters) used, the predictive map generated from the best model, and the path where data was saved.

jscavetta95/maxentools documentation built on Dec. 8, 2019, 1:26 p.m.