XGBoost gradient-boosting backend for xplus

Gradient-boosted trees via xgboost::xgb.train(). Binary targets use binary:logistic, fractional pseudo-labels reg:squarederror (clipped to [0, 1] on prediction). Hyperparameters use parsnip-harmonized names so a boost_tree() spec can map onto them later. Boosting memorizes training targets like any high-capacity learner, so fit_predict returns out-of-fold predictions computed with xgb.cv on the folds xplus() already assigned — the same honest-score role ranger's OOB plays, at the cost of one extra cross-validation pass per final fit.

Usage

xplus_backend_xgboost(
  trees = 100,
  learn_rate = 0.3,
  tree_depth = 6,
  min_n = 1,
  mtry = NULL,
  sample_size = 1,
  loss_reduction = 0,
  num.threads = 1,
  seed = NULL
)

Arguments

trees

Number of boosting rounds (xgboost nrounds).

learn_rate

Shrinkage step (xgboost eta).

tree_depth

Maximum tree depth (xgboost max_depth).

min_n

Minimum child weight (xgboost min_child_weight).

mtry

Number of features sampled per split; NULL uses all features. Converted to colsample_bynode as a proportion of ncol(x).

sample_size

Row subsampling fraction per tree (xgboost subsample).

loss_reduction

Minimum loss reduction to split (xgboost gamma).

num.threads

OpenMP threads (xgboost nthread).

seed

Optional seed passed to each fit; NULL draws from R's RNG per fit so that xplus(seed = ...) remains reproducible.

Value

An "xplus_backend" object.

See also

xplus_backend(), xplus(), xplus_backend_ranger()

Examples

if (FALSE) { # \dontrun{
fit <- xplus(x, y, backend = xplus_backend_xgboost(trees = 200))
} # }