tune_grid_h2o: Tune h2o models

In stevenpawley/h2oparsnip: Model Wrappers for H2O models

Tune h2o models

Description

This is a prototype of a version of tune_grid that uses h2o.grid to perform hyperparameter tuning.

Usage

tune_grid_h2o(
  object,
  preprocessor = NULL,
  resamples,
  param_info = NULL,
  grid = 10,
  metrics = NULL,
  control = control_h2o(),
  ...
)

Arguments

object	A parsnip 'model_spec' object.
preprocessor	A 'recipe' object.
resamples	An 'rset' object.
param_info	A 'dials::parameters()' object or NULL. If none is given, a parameters set is derived from other arguments. Passing this argument can be useful when parameter ranges need to be customized.
grid	A 'data.frame' of tuning combinations or a positive integer. The data frame should have columns for each parameter being tuned and rows for tuning parameter candidates. An integer denotes the number of candidate parameter sets to be created automatically. If a positive integer is used or no tuning grid is supplied, then a semi-random grid via 'dials::grid_latin_hypercube' is created based on the specified number of tuning iterations (default size = 10).
metrics	A 'yardstick::metric_set' or NULL. Note that not all yardstick metrics can be used with 'tune_grid_h2o'. The metrics must be one of 'yardstick::rsq', 'yardstick::rmse' or 'h2oparsnip::mse' for regression models, and 'yardstick::accuracy', 'yardstick::mn_log_loss', 'yardstick::roc_auc' or 'yardstick::pr_auc' for classification models. If NULL then the default is 'yardstick::rsq' for regression models and 'yardstick::mn_log_loss' for classification models.
control	An object used to modify the tuning process.
...	Not currently used.

Limitations

- Only model arguments can be tuned, not arguments in the preprocessing recipes.

- Parsnip only allows 'data.frame' and 'tbl_spark' objects to be passed to the 'fit' method, not 'H2OFrame' objects.

stevenpawley/h2oparsnip documentation built on June 20, 2022, 12:48 p.m.