local_simplify: Simplify maxent model

In HMB3/nenswniche: Rapidly estimate species ranges and intersect with other layers

Simplify maxent model

Description

This is a local version of rmaxent::simplify (https://github.com/johnbaums/rmaxent) Given a candidate set of predictor variables, this function identifies a subset that meets specified multicollinearity criteria. Subsequently, backward stepwise variable selection is used to iteratively drop the variable that contributes least to the model, until the contribution of each variable meets a specified minimum, or until a predetermined minimum number of predictors remains.

It assumes that the input df is that returned by the fit_maxent_targ_bg_back_sel function

Usage

local_simplify(
  occ,
  bg,
  path,
  taxa_column = "taxa",
  response_curves = TRUE,
  logistic_format = TRUE,
  type = "PI",
  cor_thr,
  pct_thr,
  k_thr,
  features = "lpq",
  replicates = 1,
  quiet = TRUE
)

Arguments

occ	SpatialPointsDataFrame - Spdf of all taxa records returned by the 'prepare_sdm_table' function
bg	SpatialPointsDataFrame - Spdf of of candidate background points
path	Character string - Vector of enviro conditions that you want to include
taxa_column	Character string - Vector of enviro conditions that you want to include
logistic_format	Logical value indicating whether maxentResults.csv should report logistic value thresholds
type	The variable contribution metric to use when dropping variables
cor_thr	Numeric - The max allowable pairwise correlation between predictor variables
pct_thr	Numeric - The min allowable percent variable contribution
k_thr	Numeric - The min number of variables to be kept in the model
features	Character string - Which features should be used? (e.g. linear, product, quadratic 'lpq')
replicates	Numeric - The number of replicates to use
responsecurves	Logical - Save response curves of the maxent models (T/F)?

HMB3/nenswniche documentation built on Jan. 31, 2023, 11:46 p.m.