compare: Compare multiple learners with optional tuning.

In funcml: Functional Machine Learning Framework

Compare multiple learners with optional tuning.

Description

Compare multiple learners with optional tuning.

Usage

compare(
  data,
  formula,
  models,
  specs = NULL,
  resampling = cv(5),
  metrics = NULL,
  type = NULL,
  conf_level = 0.95,
  seed = NULL,
  ncores = NULL,
  tune = FALSE,
  grids = NULL,
  metric = NULL,
  ...
)

Arguments

data	Data frame.
formula	Model formula.
models	Character vector of learner ids.
specs	Optional named list of fixed specs per learner.
resampling	Resampling object from cv().
metrics	Character vector of metrics to report. When tune = TRUE, these are computed for each learner's tuned best configuration.
type	Prediction type override.
conf_level	Confidence level for learner summary intervals.
seed	Optional seed.
ncores	Optional number of CPU cores used to compare learners. NULL or 1 runs sequentially.
tune	Logical; if TRUE, run tune() for each learner before comparing.
grids	Optional tuning grids. Supply either a single data frame to reuse across learners or a named list of data frames keyed by learner id.
metric	Optimization metric used when tune = TRUE.
...	Additional arguments passed to evaluate() or tune() / fit().

Value

A funcml_compare object.

Examples

cmp <- compare(
  data = mtcars,
  formula = mpg ~ wt + hp,
  models = c("glm", "rpart"),
  resampling = cv(3, seed = 1),
  metrics = c("rmse", "mae")
)
cmp$results

funcml documentation built on Aug. 22, 2026, 5:08 p.m.