Skip to contents

Assign trials to train or test sets

assign_sample_sets.Rd

Randomly assigns each non-catch trial to either a training set or a test set using a stratified split within each participant. Catch trials (sampleAlg == "check") receive NA and are excluded from both sets.

Usage

assign_sample_sets(df, test_prop = 0.2, seed = 42)

Arguments

df

Data frame with columns worker_id and sampleAlg. sampleAlg must contain the values "check", "random", and/or "validation".

test_prop

Numeric between 0 and 1. Proportion of non-check trials to assign to the test set. Default: 0.2.

seed

Integer. Random seed passed to set.seed for reproducibility. Default: 42.

Value

The input data frame with an additional character column sampleSet containing "train", "test", or NA (for catch trials).

Details

The split is performed per participant so that each participant contributes approximately test_prop of their trials to the test set. Setting seed ensures the assignment is reproducible.

Examples

if (FALSE) { # \dontrun{
d <- data.frame(
  worker_id = rep("p1", 10),
  sampleAlg = c(rep("random", 8), rep("check", 2))
)
assign_sample_sets(d, test_prop = 0.2, seed = 1)
} # }