LEGIT (version 1.4.1)

longitudinal_folds: Longitudinal folds

Description

Function to create folds adequately for longitudinal datasets by forcing every observation with the same id to be in the same fold. Can be used with LEGIT_cv to make sure that the cross-validation folds are appropriate when using longitudinal data.

Usage

longitudinal_folds(
  cv_iter = 1,
  cv_folds = 10,
  id,
  formula = NULL,
  data = NULL,
  data_needed = NULL,
  print = TRUE
)

Value

Returns a list of vectors containing the fold number for each observation

Arguments

cv_iter

Number of cross-validation iterations (Default = 1).

cv_folds

Number of cross-validation folds (Default = 10).

id

Factor vector containing the id number of each observation.

formula

Optional Model formula. If data and formula are provided, only the non-missing observations will be used when creating the folds (Put "formula" here if you have missing data).

data

Optional data.frame used for the formula. If data and formula are provided, only the non-missing observations will be used when creating the folds (Put "data" here if you have missing data).

data_needed

Optional data.frame with variables that have to be included (Put "cbind(genes,env)"" or "latent_var" here if you have missing data).

print

If FALSE, nothing except warnings will be printed. (Default = TRUE).

Examples

train = example_2way(500, 1, seed=777)
# Assuming it's longitudinal with 4 timepoints, even though it's not
id = factor(rep(1:125,each=4))
fit_cv = LEGIT_cv(train$data, train$G, train$E, y ~ G*E, folds=longitudinal_folds(1,10, id))