step for dummy variables across multiple columns

[topepo]

In cases where a row of a data set can have multiple categories spread across columns, a step to find the unique set of categories, then create dummy columns for each. Each row can have multiple 1's in the new dummy variables.

[EmilHvitfeldt]

I can take this one! step_multi_dummy()?

NA should be counted as none right?

[topepo]

I think that step_dummy_multi_label() (for tab-complete) would be a good idea.

NA should be counted as none right?

Correct.

I'm not sure if there is an easy way to define the prefix for the new indicator variables (from the data).

Just so we are clear, imagine columns for language spoken. There data might look like:

lang_1     lang_2     lang_3
English    Italian    NA
Spanish    NA         French
Armenian   English    French
NA         NA         NA

We would end up with

Armenian English French Italian Spanish
       0       1      0       1       0
       0       0      1       0       1
       1       1      1       0       0
       0       0      0       0       0

[EmilHvitfeldt]

I think that step_dummy_multi_label() (for tab-complete) would be a good idea.

Perfect.

I think a good default for the prefix would be based on the first column selected.

Should the step have a way to deal with a high number of labels? I was thinking of a threshold argument like step_other() could be useful. Defaulting to 0. You could always step_other() each individual column, but It would be nice to do it globally as well.

[topepo]

I think that adding a threshold argument for an other category would be a great idea.

[github-actions]

This issue 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.