Keep original labels in step_discretize

[EmilHvitfeldt]

This PR aims to close #674.

It does 2 things

• makes it so when prefix = NULL is set in discretize(), that the factor levels match what happens in cut(). It essentially disable the custom labeling.
• Sets the default value of prefix in discretize() to NULL.

The old behavior is still possible by setting prefix to "bin". It think that changing the default is worthwhile since it aligns it better with what happens in the {embed} steps.

library(recipes)
library(modeldata)
data(biomass)

recipe(HHV ~ carbon, data = biomass) %>% 
  step_discretize(carbon) %>%
  prep() %>%
  bake(new_data = NULL)
#> # A tibble: 536 × 2
#>    carbon        HHV
#>    <fct>       <dbl>
#>  1 (49.7, Inf]  20.0
#>  2 (47.1,49.7]  19.2
#>  3 (47.1,49.7]  18.3
#>  4 (44.7,47.1]  18.2
#>  5 (44.7,47.1]  18.4
#>  6 (44.7,47.1]  18.5
#>  7 (47.1,49.7]  18.7
#>  8 (44.7,47.1]  18.3
#>  9 (47.1,49.7]  18.6
#> 10 (44.7,47.1]  18.9
#> # … with 526 more rows

recipe(HHV ~ carbon, data = biomass) %>% 
  step_discretize(carbon, options = list(prefix = "bin")) %>%
  prep() %>%
  bake(new_data = NULL)
#> # A tibble: 536 × 2
#>    carbon   HHV
#>    <fct>  <dbl>
#>  1 bin4    20.0
#>  2 bin3    19.2
#>  3 bin3    18.3
#>  4 bin2    18.2
#>  5 bin2    18.4
#>  6 bin2    18.5
#>  7 bin3    18.7
#>  8 bin2    18.3
#>  9 bin3    18.6
#> 10 bin2    18.9
#> # … with 526 more rows

^{Created on 2022-04-13 by the reprex package (v2.0.1)}

[topepo]

This should also add a keep_original_cols and error if the prefix is NULL

[EmilHvitfeldt]

Which function should error if prefix = NULL?

[topepo]

if (keep_original_cols && is.null(prefix), the bake method should error.

[EmilHvitfeldt]

I don't think that keep_original_cols fits for this step. I wrote up something a little longer here #929 (comment). In essence; so far we only used keep_original_cols in steps that generate new columns. This step doesn't do that.

[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.