shuffle_grouped_data {designit}

Generate in one go a shuffling function that produces permutations with specific constraints on multiple sample variables and group sizes fitting one specific allocation variable

Description

Generate in one go a shuffling function that produces permutations with specific constraints on multiple sample variables and group sizes fitting one specific allocation variable

Usage

shuffle_grouped_data(
  batch_container,
  allocate_var,
  keep_together_vars = c(),
  keep_separate_vars = c(),
  n_min = NA,
  n_max = NA,
  n_ideal = NA,
  subgroup_var_name = NULL,
  report_grouping_as_attribute = FALSE,
  prefer_big_groups = FALSE,
  strict = TRUE,
  fullTree = FALSE,
  maxCalls = 1e+06
)

Arguments

batch_container

Batch container with all samples assigned that are to be grouped and sub-grouped

allocate_var

Name of a variable in the samples table to inform possible groupings, as (sub)group sizes must add up to the correct totals

keep_together_vars

Vector of column names in sample table; groups are formed by pooling samples with identical values of all those variables

keep_separate_vars

Vector of column names in sample table; items with identical values in those variables will not be put into the same subgroup if at all possible

n_min

Minimal number of samples in one sub(!)group; by default 1

n_max

Maximal number of samples in one sub(!)group; by default the size of the biggest group

n_ideal

Ideal number of samples in one sub(!)group; by default the floor or ceiling of mean(n_min,n_max), depending on the setting of prefer_big_groups

subgroup_var_name

An optional column name for the subgroups which are formed (or NULL)

report_grouping_as_attribute

Boolean, if TRUE, add an attribute table to the permutation functions' output, to be used in scoring during the design optimization

prefer_big_groups

Boolean; indicating whether or not bigger subgroups should be preferred in case of several possibilities

strict

Boolean; if TRUE, subgroup size constraints have to be met strictly, implying the possibility of finding no solution at all

fullTree

Boolean: Enforce full search of the possibility tree, independent of the value of maxCalls

maxCalls

Maximum number of recursive calls in the search tree, to avoid long run times with very large trees

Value

Shuffling function that on each call returns an index vector for a valid sample permutation