Step_Novel Use & Configuration

tidyverse
1

this is a follow-up question to an excellent post step_other Vs. step_novel Vs. step_unknown - #4 by yyu.

First, if new levels are replaced with a value of "new" in recipe step_novel how does a model predict against records with levels called "new" that have never been seen? Are these skipped, omitted or treated as NA or handled differently by different types of model engines?

Second, I see that there is an argument allow_novel_levels in two places of the final model object:

m[["pre"]][["actions"]][["recipe"]][["blueprint"]][["allow_novel_levels"]]
m[["pre"]][["mold"]][["blueprint"]][["allow_novel_levels"]]

My question is what's the difference between these arguments and which one should I set to TRUE so new levels are set as "new" in my prediction data? Then, how do I view the transformed (baked?) prediction data results showing the "new" attribute levels.

thank you

2

This topic was automatically closed 21 days after the last reply. New replies are no longer allowed.

If you have a query related to it or one of the replies, start a new topic and refer back with a link.