cv_split_group_kfold: Resamples for group K-fold cross-validation with...

In statist-bhfz/resampleR: Easy resampling with data.table

Description

Resamples for group K-fold cross-validation with stratification by mean value of target variable.

Usage

cv_split_group_kfold(data, y, id, nfolds = 5L, probs = seq(0, 1,
  length.out = 11))

Arguments

data	data.table with y and id.
y	Target variable name (character).
id	Identifier of each group of observations (character).
nfolds	Number of folds (min 2, max 20).
probs	Numeric vector of probabilities for quantile binning with values in [0, 1] range.

Details

Numeric target: quantile binning is used for stratification.

Character/categorical target: resampling performs within categories.

probs can be a vector like c(0, seq(0.99, 1, length.out = 10)) for target with very skewed distribution, e.g. for financial data with 99% of 0's.

Ensures that all observations for each id will be placed in the same fold.

Value

data.table with nfolds columns. Each column is an indicator variable with 1 corresponds to observations in validation dataset (stratified by target).

Examples

dt <- data.table(
    user = rep(1:100, each = 5),
    target = rnorm(5e2)
)
cv_split_group_kfold(dt, "target", "user")

statist-bhfz/resampleR documentation built on Sept. 2, 2019, 8:14 p.m.