EN_model: EN MODELLING FUNCTION

In agarciaEE/NINA: Niche Interaction Network Analyses

EN MODELLING FUNCTION

Description

Samples pseudo-absences from a environmental extent based on euclidean distances to the observations in a environmental PCA

Usage

EN_model(
  env,
  occ,
  res = NULL,
  path = "./",
  project.name = "NINA_EN",
  nstart = 25,
  k.max = NULL,
  B = 100,
  crs = "+proj=longlat +datum=WGS84 +ellps=WGS84 +towgs84=0,0,0",
  extrapolate.niche = FALSE,
  save.bootstraps = F,
  save.model = F,
  cor = F,
  relative.niche = T,
  h = "href",
  mask = NULL,
  th.o = NULL,
  th.s = NULL,
  density.method = c("epa", "bivnorm"),
  combine.clusters = FALSE,
  cluster = NULL,
  n.clus = NULL,
  R = 100,
  sample.pseudoabsences = TRUE,
  eval = FALSE,
  split.data = FALSE,
  split.percentage = 0.25,
  split.method = c("kmeans", "Euclidean"),
  plot.eval = FALSE,
  rep = 100,
  th = NULL,
  ras = NULL,
  best.th = c("accuracy", "similarity"),
  bootstraps = 1,
  assemble.models = TRUE,
  assemble.method = c("ACC", "Jaccard Similarity", "TSS", "AUC", "kappa")
)

Arguments

env	Environmental data frame
occ	Occurrence dataset
res	grid resolution of the spatial extent
path	Directory path of partition models
project.name	String indicating the project name. Default is "NINA_EN"
nstart	Burn up start for clustering estimation
k.max	Maximum number of cluseters
B	Number of runs
crs	CRS object or a character string describing a projection and datum in PROJ.4 format
extrapolate.niche	Boolean wheter to allow niche extrapolation if needed
save.bootstraps	Boolean whether to save partitions
save.model	Boolean whether to save the model
cor	Logical
relative.niche	logical. Only in case of using clustering method. If TRUE, computes the relative species niche density over the overall species niche clusters.
h	smoothing parameter for the kernel estimation. Default is 'href'. Altenrtanively can be set to 'LSCV' or any given numeric value
mask	raster mask to resample the created kernel densiity grid raster
th.o	numeric threshold to filter density values of occurrences
th.s	numeric threshold to filter density values of environment
density.method	"epanechnikov" or "bivnorm"
combine.clusters	Boolean whether to combine regions into global models
cluster	data frame with clustered regions as levels
n.clus	number of clusters to perform
R	niche space grid
sample.pseudoabsences	Boolean to whether sample pseudo-absences
eval	Boolean whether to evaluate the model
split.data	Boolean whether to split data in case of model bootstrapping
split.percentage	Split percentage, Default is 0.25
split.method	String indicating the method to split occurrence data. Default is kmeans
plot.eval	Logical to whether plot the evaluation
rep	number of randomzation tests
th	threshold to perform cut off for model evaluation
ras	raster to constrain pseudoabsences sampling in model evalluation
best.th	method to select the best thresholt. Default is "similarity"
bootstraps	Integer indicating number of partitions to perform. Default is 1
assemble.models	Boolean whether to assemble all partitions into ensemble model
assemble.method	String indicating the evaluation parameter to weight and compute the assembling. Default is "ACC"

Details

Returns an error if filename does not exist.

Value

List of elements

Examples

## Not run: 
EN<- EN_model(env_data, occ_data1, boot)

## End(Not run)
## Not run: 
EN<- EN_model(env_data, occ_data2, cluster = "env", n.clus = 5)

## End(Not run)

agarciaEE/NINA documentation built on Jan. 9, 2025, 10:09 a.m.