Run Bayesian Optimization Benchmark — run_bayesian

Convenience function to run Bayesian-optimized learners in nested CV and compare performance.

Usage

run_bayesian_benchmark(
  task,
  learners = c("xgboost", "ranger", "glmnet"),
  outer_folds = 5,
  inner_folds = 3,
  n_evals = NULL,
  measure = "classif.auc",
  seed = NULL
)

Arguments

task: An mlr3 Task
learners: Character vector of learner names: "xgboost", "lightgbm", "ranger", "glmnet"
outer_folds: Number of outer CV folds (default: 5)
inner_folds: Number of inner CV folds (default: 3)
n_evals: Number of MBO iterations per learner (default: auto)
measure: Performance measure (default: classif.auc)
seed: Random seed

Value

A benchmark result with tuned learners

Examples

if (FALSE) { # \dontrun{
if (requireNamespace("mlr3mbo", quietly = TRUE) &&
    requireNamespace("mlr3learners", quietly = TRUE) &&
    requireNamespace("xgboost", quietly = TRUE) &&
    requireNamespace("ranger", quietly = TRUE) &&
    requireNamespace("glmnet", quietly = TRUE)) {
  result <- run_bayesian_benchmark(
    task,
    learners = c("xgboost", "ranger", "glmnet"),
    outer_folds = 5,
    n_evals = 30
  )
  result$aggregate(msr("classif.auc"))
}
} # }