create_folds_mc: Creation of folds for group-out cross-validation based on...

In DrrDom/pfpp: Different convenience functions (file loading, caret model processing, etc)

Creation of folds for group-out cross-validation based on Monte Carlo method.

Description

Function takes input vector of variables which will be used to group input objects. It creates folds with approaximately equal number of objects in each fold. All object of the same group put in the single fold.

Usage

create_folds_mc(
  v,
  nfolds = "auto",
  max_iter = 1000,
  start_opt_param = 100,
  error_limit = NA,
  seed = 0
)

Arguments

v	input vector of group variables.
nfolds	maximum number of folds. But if it is not possible to create the specified number of folds its number will be automatically decreased.
max_iter	maximum number of Monte Carlo iterations.
start_opt_param	strting parameter of Monter Carlo optimization.
error_limit	if final error will be greater than this values results will be discarded.
seed	seed for random number generator to reproduce results.

Value

vector of folds numbers for each input object.

Examples

df <- data.frame(A=runif(100), B=runif(100), C=sample(LETTERS[1:10], 100, TRUE))
res <- create_folds_mc(df$C, 5)

DrrDom/pfpp documentation built on April 17, 2024, 10:24 a.m.