Skip to contents

Run the simulation workflows used in the article and thesis.

Usage

run_simulation_study(
  dimensions = list(c(500L, 100L), c(500L, 20L), c(100L, 20L), c(80L, 25L), c(60L, 33L),
    c(40L, 50L), c(20L, 100L)),
  true_ncomp = c(2L, 4L, 6L),
  missing_props = seq(5, 50, 5),
  mechanisms = c("MCAR", "MAR"),
  reps = 1L,
  seed = NULL,
  max_ncomp = 8L,
  criteria = c("Q2-LOO", "Q2-10fold", "AIC", "AIC-DoF", "BIC", "BIC-DoF"),
  incomplete_methods = c("nipals_standard", "nipals_adaptative"),
  imputation_methods = c("mice", "knn", "svd"),
  folds = 10L,
  mar_y_bias = 0.8
)

Arguments

dimensions

List of (n, p) integer pairs.

true_ncomp

Vector of true component counts.

missing_props

Missing-data proportions as fractions or percentages.

mechanisms

Missing-data mechanisms.

reps

Number of replicates.

seed

Optional base random seed.

max_ncomp

Maximum number of extracted components.

criteria

Criteria evaluated on complete and imputed data.

incomplete_methods

Incomplete-data NIPALS workflows.

imputation_methods

Imputation methods.

folds

Number of folds used by "Q2-10fold".

mar_y_bias

MAR bias parameter passed to add_missingness().

Value

A data frame with one row per study run.