Clarify discussion of strata

[juliasilge]

This PR clarifies how strata works, especially with regards to numeric stratification variables. It uses templating to minimize repetition.

[hfrick]

the reduction of repetition is rather nice 😍

[samuelmacedo83]

Hi @juliasilge,
I was reading the documentation about make_strata and vfold_cv and I'm a lit bit confuse.
In make_strata, the documentation said that numeric variables are divided by percentile, but in vfold_cv the numeric variables are divided by quartile.

So, even if I use, for example, a 10 fold cv, the numeric variable will be divided by quartiles?

[juliasilge]

The function make_strata() uses the stats::quantile() function with a default of breaks = 4, which means quartiles:

rsample/R/make_strata.R

Line 119 in 62902cb

pctls <- quantile(x, probs = (0:breaks) / breaks, na.rm = TRUE)

You can create different bins (e.g. use different percentiles of the data) by changing breaks. Maybe we can clarify the make_strata() documentation a bit more.

[samuelmacedo83]

I think I understood. So the breaks parameter on vfold_cv is related to how to stratify the variable defined on strata parameter.
It makes sense :)
thank you @juliasilge

[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.