mlh_outsample_row_indices {kdry}

mlh_outsample_row_indices

Description

Machine learning helper function to convert a vector of (in- sample) row indices of a fold into out-of-sample row indices.

Usage

mlh_outsample_row_indices(fold_list, dataset_nrows, type = NULL)

Arguments

fold_list

A list of integer vectors that describe the row indices of cross-validation folds. The list must be named.

dataset_nrows

An integer. The number of rows in the dataset dataset. This parameter is required in order to compute the out-of-sample row indices.

type

A character. To be used if the out-of-sample row indices need to be formatted in a special manner (default: NULL). Currently, the only allowed value is type = "glmnet" in order to format the row indices as required by glmnet::cv.glmnet's argument foldid.

Value

If type = NULL, returns a list of same length as fold_list with each item containing a vector of out-of-sample row indices. If type = "glmnet", a data.table is returned with two columns and each row representing one observation of the dataset that is assigned to a specific test fold. The column "fold_id" should be passed further on to the argument foldid of glmnet::cv.glmnet.

Examples

fold_list <- list(
  "Fold1" = setdiff(seq_len(100), 1:33),
  "Fold2" = setdiff(seq_len(100),66:100),
  "Fold3" = setdiff(seq_len(100),34:65)
)
mlh_outsample_row_indices(fold_list, 100)
mlh_outsample_row_indices(fold_list, 100, "glmnet")