select_one_sample_per: Random sample one case per combination of factors

In njjms/nicks: Basic Functions for Biostatistics

Random sample one case per combination of factors

Description

To prevent dependence between samples when repeated measures are used.

Usage

select_one_sample_per(.data, seed = 100, ...)

Arguments

.data	a dataframe
seed	random seed by default set to 100
...	grouping columns (e.g. patient ID, confirmed biopsy truth status, etc.)

Value

an ungrouped dataframe containing only the randomly selected rows

Examples

nrows <- 300
test_data <- data.frame(
	patient_ids = sort(sample(1:100, nrows, replace=TRUE)),
	test = runif(n=nrows),
	casefile_id = seq_along(1:nrows)
)

test_data %>%
	mutate(
		status = if_else(test < .80, "STABLE", "AR")
	) %>%
	select(-test) -> test_data

select_one_sample_per(test_data, patient_ids, status)

njjms/nicks documentation built on May 4, 2022, 8:10 a.m.