Update vfold.R

[ZWael]

Hello,
this pull request is for bug correction with vfold_splits function
actually the vfold_splits function do not forward the breaks argument to vfold_splits call

Symptoms
the de fault breaks = 4 is used instead

library(tidymodels)
set.seed(123)
d=iris[1:50,]
folds=vfold_cv(d, v = 5, repeats = 2,strata = Petal.Width)
#> Warning: The number of observations in each quantile is below the recommended threshold of 20.
#> • Stratification will use 2 breaks instead.
#> The number of observations in each quantile is below the recommended threshold of 20.
#> • Stratification will use 2 breaks instead.
folds=vfold_cv(d, v = 5, repeats = 2,strata = Petal.Width,
               breaks = 2)
#> Warning: The number of observations in each quantile is below the recommended threshold of 20.
#> • Stratification will use 2 breaks instead.
#> The number of observations in each quantile is below the recommended threshold of 20.
#> • Stratification will use 2 breaks instead.

obviously not forwarding the breaks argument

no error after vfold_splits call modification

^{Created on 2024-04-12 with reprex v2.1.0}

[ZWael]

Hello @hfrick,
Could you check my bug fix ?

[hfrick]

Thanks for the indirect bug report and the PR!

Here's the slightly reworked reprex for my own reference.

With this PR:

library(rsample)

iris_small <- iris[1:50,]

set.seed(123)

# default breaks = 4
folds <- vfold_cv(iris_small, v = 5, repeats = 2, strata = Petal.Width)
#> Warning: The number of observations in each quantile is below the recommended threshold of 20.
#> • Stratification will use 2 breaks instead.
#> The number of observations in each quantile is below the recommended threshold of 20.
#> • Stratification will use 2 breaks instead.

# this should not warn anymore
folds <- vfold_cv(iris_small, v = 5, repeats = 2, strata = Petal.Width, breaks = 2)

^{Created on 2024-04-22 with reprex v2.1.0}

[github-actions]

This pull request has been automatically locked. If you believe you have found a related problem, please file a new issue (with a reprex: https://reprex.tidyverse.org) and link to this issue.