create_groups {stressor}

Create Groups for CV

Description

Create groups for the data by separating them either into 10 fold cross-validation, LOO cross-validation, or k-means grouping.

Usage

create_groups(
  formula,
  data,
  n_folds = 10,
  k_mult = NULL,
  repl = FALSE,
  grouping_formula = NULL
)

Arguments

formula

A formula object that specifies the model to be fit.

data

The data that will be separated into each group.

n_folds

An integer value defaulted to 10 fold cross-validation. If NULL uses Leave One Out(LOO) instead.

k_mult

When specified, this is passed onto the cv_cluster to fit the data into k_groups.

repl

A Boolean value defaulted to 'FALSE', change to 'TRUE' when replicates need to be included in the same group.

grouping_formula

A formula object that specifies how the groups will be gathered.

Details

If 'k_mult' is specified as an integer, the formula object will be used to help determine the features specified by the user. This will be passed to the cv_cluster function, which takes a scaled matrix of features.

This function is called by the cv methods as it forms the groups necessary to perform the cross-validation. If you want to use this, it is a nice function that separates the 'data' into groups for training and testing.

Value

A vector of the length equal to number of rows of data.frame from the data argument.

Examples

 # data generation
 lm_data <- data_gen_lm(1000)

 # 10 Fold CV group
 create_groups(Y ~ ., lm_data)

 # Spatial CV
 create_groups(Y ~ ., lm_data, n_folds = 10, k_mult = 5)

 # LOO CV group
 create_groups(Y ~ ., lm_data, n_folds = NULL)