Learn R Programming

dials

Overview

This package contains infrastructure to create and manage values of tuning parameters for the tidymodels packages. If you are looking for how to tune parameters in tidymodels, please look at the tune package and tidymodels.org.

The name reflects the idea that tuning predictive models can be like turning a set of dials on a complex machine under duress.

Installation

You can install the released version of dials from CRAN with:

install.packages("dials")

You can install the development version from Github with:

# install.packages("pak")
pak::pak("tidymodels/dials")

Contributing

Please note that the dials project is released with a Contributor Code of Conduct. By contributing to this project, you agree to abide by its terms.

Copy Link

Version

Install

install.packages('dials')

Monthly Downloads

26,978

Version

1.4.5

License

MIT + file LICENSE

Issues

Pull Requests

Stars

Forks

Maintainer

Hannah Frick

Last Published

November 11th, 2026

Functions in dials (1.4.5)

freq_cut

Near-zero variance parameters
harmonic_frequency

Harmonic Frequency
dropout

Neural network parameters
get_batch_sizes

Get batch sizes
mtry

Number of randomly sampled predictors
grid_regular

Create grids of tuning parameters
min_dist

Parameter for the effective minimum distance between embedded points
learn_rate

Learning rate
num_leaves

Possible engine parameters for lightbgm
max_tokens

Maximum number of retained tokens
max_times

Word frequencies for removal
num_runs

Number of Computation Runs
min_unique

Number of unique values for pre-processing
num_tokens

Parameter to determine number of tokens in ngram
momentum

Gradient descent momentum parameter
ordinal_link

Ordinal regression link functions (character)
mixture

Mixture of penalization terms
over_ratio

Parameters for class-imbalance sampling
parallel_reg

Parallel regression assumption (logical)
num_clusters

Number of Clusters
parameters

Create a parameter set
num_breaks

Number of cut-points for binning
predictor_prop

Proportion of predictors
prior_slab_dispersion

Bayesian PCA parameters
parameters_constr

Construct a new parameter set object
penalty

Amount of regularization/penalization
scheduler-param

Parameters for neural network learning rate schedulers
neighbors

Number of neighbors
rln-param

Parameters for regularization learning networks
prune_method

MARS pruning methods
num_comp

Number of new features
max_nodes

Parameters for possible engine parameters for randomForest
mtry_prop

Proportion of Randomly Selected Predictors
range_limits

Limits for the range of predictions
new-param

Tools for creating new parameter objects
regularization_method

Estimation methods for regularized models
num_knots

Number of knots (integer)
prop_terms

Proportion of top predictors
rbf_sigma

Kernel parameters
stop_iter

Early stopping parameter
regularization_factor

Parameters for possible engine parameters for ranger
range_validate

Tools for working with parameter ranges
smoothness

Kernel Smoothness
target_weight

Amount of supervision parameter
num_hash

Text hashing parameters
select_features

Parameter to enable feature selection
vocabulary_size

Number of tokens in vocabulary
weight

Parameter for "double normalization" when creating token counts
validation_set_prop

Proportion of data used for validation
value_validate

Tools for working with parameter values
resnet-param

Parameters for residual networks
surv_dist

Parametric distributions for censored data
scale_pos_weight

Parameters for possible engine parameters for xgboost
summary_stat

Rolling summary statistic for moving windows
window_size

Parameter for the moving window size
weight_func

Kernel functions for distance weighting
threshold_structure

Threshold structure (character)
reexports

Objects exported from other packages
weight_scheme

Term frequency weighting methods
update.parameters

Update a single parameter in a parameter set
unknown

Placeholder for unknown parameter values
threshold

General thresholding parameter
tab-pfn-param

Parameters for TabPFN models
survival_link

Survival Model Link Function
shrinkage_correlation

Parameters for possible engine parameters for sda models
token

Token types
trees

Parameter functions related to tree- and rule-based models.
trim_amount

Amount of Trimming
type_sum.param

Succinct summary of parameter objects
all_neighbors

Parameter to determine which neighbors to use
l2_leaf_reg

Possible engine parameters for catboost
activation

Activation functions between network layers
Laplace

Laplace correction parameter
confidence_factor

Parameters for possible engine parameters for C5.0
bart-param

Parameters for BART models These parameters are used for constructing Bayesian adaptive regression tree (BART) models.
cal_method_class

Methods for model calibration
attention-param

Parameters for attention-based tabular models
buffer

Buffer size
adjust_deg_free

Parameters to adjust effective degrees of freedom
dials-package

dials: Tools for working with tuning parameters
class_weights

Parameters for class weights for imbalanced problems
extrapolation

Parameters for possible engine parameters for Cubist
cost

Support vector machine parameters
conditional_min_criterion

Parameters for possible engine parameters for partykit models
deg_free

Degrees of freedom (integer)
degree

Parameters for exponents
max_num_terms

Parameters for possible engine parameters for earth models
grid_hybrid

Hybrid space-filling/regular parameter grids
dist_power

Minkowski distance parameter
initial_umap

Initialization method for UMAP
grid_space_filling

Space-filling parameter grids
encode_unit

Class for converting parameter values back and forth to the unit range
finalize

Functions to finalize data-specific parameter ranges
grid_max_entropy

Max-entropy and latin hypercube grids