subsampling: Subsampling permutations of clustering dataset

In vittoriofortino84/COPS: Clustering algorithms for Omics-driven Patient Stratification

Subsampling permutations of clustering dataset

Description

Creates subsamples via cross-validation folds or bootstrapping.

Usage

subsampling(..., subsampling_strategy = "cv")

cv_fold(
  dat_list,
  nfolds = 5,
  nruns = 2,
  stratified_cv = FALSE,
  anti_stratified = FALSE,
  cv_stratification_var = NULL,
  extra_fold = TRUE,
  ...
)

bootstrap(dat_list, nruns = 100, ...)

Arguments

...	extra arguments are ignored
subsampling_strategy	either "cv" or "bootstrap" (not implemented)
dat_list	list of datasets, each either data.table or data.frame (samples x features) with an "id" column or expression matrix (genes x samples) with named columns
nfolds	number of cross-validation folds
nruns	number of cross-validation replicates
stratified_cv	if TRUE, perform stratified sampling for folds
anti_stratified	if TRUE, maximize separation of batch labels within folds, opposite of stratified sampling
cv_stratification_var	labels used for stratification
extra_fold	if TRUE, generates an extra fold (nfolds+1) that corresponds to the full dataset

Value

list of data.frames with added columns "fold", "run" and "cv_index" as well as duplicated rows of the original data corresponding to different folds.

list of data.frames with added columns "fold", "run" and "cv_index" as well as duplicated rows of the original data corresponding to different folds.

list of data.frames

Functions

• cv_fold(): Cross-validation based subsampling

• bootstrap(): Subsampling via bootstrapping.

vittoriofortino84/COPS documentation built on Jan. 28, 2025, 3:16 p.m.