mlr3 Resampling for BORG Cross-Validation

Creates a native mlr3::Resampling object from BORG cross-validation folds, enabling direct use with mlr3 benchmarking infrastructure.

Usage

borg_to_mlr3(data = NULL, folds = NULL, cv_obj = NULL, id = "borg")

Arguments

data

A data frame. Required when using folds directly, not needed when using cv_obj.

folds

A list of lists, each with train and test integer index vectors.

cv_obj

A borg_cv object. If provided, folds are extracted from it.

id

Character. Resampling identifier. Default: "borg".

Value

An mlr3::Resampling R6 object that can be used directly with mlr3::resample(), mlr3::benchmark(), and other mlr3 infrastructure.

Details

The returned resampling is pre-instantiated — it contains fixed train/test splits that respect BORG's spatial/temporal blocking. Calling $instantiate() is not needed (and will not overwrite the existing splits).

See also

borg_cv, borg_rset

Examples

# \donttest{
if (requireNamespace("mlr3", quietly = TRUE)) {
  d <- data.frame(x = runif(100), y = runif(100), z = rnorm(100))
  cv <- borg_cv(d, coords = c("x", "y"), target = "z")
  resampling <- borg_to_mlr3(cv_obj = cv)
  resampling$iters  # number of folds
}
# }