Split Indices by Variables

Extracts the row indices of a dataset, split by one or more variables. This function is useful for creating grouped analyses, stratified sampling, or preparing data for cross-validation without actually duplicating the data.

Usage

split_indices(x, ...)

# S3 method for class 'multiframe'
split_indices(x, ..., collapse = FALSE)

Arguments

x

The object to split (data frame, multidesign, multiframe, etc.)

...

Unquoted names of variables to split by

collapse

Logical; whether to collapse the resulting indices

Value

A tibble or list containing:

group variables

The splitting variables and their values

indices

List column of integer vectors with row indices for each group

Details

The function returns indices rather than the actual data, which is memory-efficient for large datasets. It works with various object types: * For data frames: Returns indices grouped by specified variables * For multidesign objects: Returns indices grouped by design variables * For multiframe objects: Returns indices grouped by design variables

The resulting indices can be used to subset the original data or to create training/testing splits for machine learning.

See also

fold_over for creating cross-validation folds, split for splitting the actual data

Examples

# With a multidesign object
mds <- multidesign(matrix(rnorm(100*10), 100, 10),
                  data.frame(group=rep(letters[1:4], each=25),
                             block=rep(1:5, times=20)))
mds_indices <- split_indices(mds, group)