tune.rfspav: Tune rfspav model (select best parameters)

In pejovic/rfspav: What the Package Does (One Line, Title Case)

Description

Do cross-validation procedured for tuning meta-parameters

Usage

tune.rfspav(
  formula,
  data_resample,
  coord_names,
  param_grid,
  metric = c("rmse", "rsq"),
  type = c("rf", "rfsp", "rfsi", "rfsig"),
  progress = TRUE,
  cpus = 1,
  importance = "impurity",
  gower_vars = NA,
  clust_style = c("sd", "equal", "pretty", "quantile", "kmeans", "hclust"),
  num_class = 10,
  feature_select = FALSE,
  data_crs = NA,
  ...
)

Arguments

formula	formula
data_resample	'rsample' folds
coord_names	PARAM_DESCRIPTION
param_grid	grid of four meta-parameters: neighbors, mtry, trees, min.node,size.
metric	Regression metrics, Default: c("rmse", "rsq")
type	Type of Random Forest Spatial Variants, Default: c("rf", "rfsp", "rfsi", "rfsig")
progress	PARAM_DESCRIPTION, Default: TRUE
cpus	Number of cores, Default: 1
importance	Variable importance, Default: 'impurity'
gower_vars	Variable used of gower-distance calculation, Default: NA
clust_style	Type of methods for choosing univariate class intervals of target variables (only for "rfsp" approach), Default: c("sd", "equal", "pretty", "quantile", "kmeans", "hclust")
num_class	Number of class intervals, Default: 10
feature_select	Should Feature Selection algorithm be applied (only for "rfsig" approach), Default: FALSE
data_crs	Data CRS, Default: NA
...	Additional parameters for 'ranger' Random Forest

Details

DETAILS

Value

Tuning list

Examples

## Not run: 
if(interactive()){
 #EXAMPLE1
 }

## End(Not run)

pejovic/rfspav documentation built on Feb. 18, 2022, 3:44 a.m.