By itself, I don't think tidymodels has this capability. But I think there are a few extension packages that interface tidymodels with a stan/brms backend.
This was very informative; thank you. I noticed a typo/inconsistency in your case_when statement (around 11:50) where you were converting TRUE/FALSE to "real"/"fake". You noted that monster_real was a character so you quoted FALSE, but you forgot to quote "TRUE". In the end it didn't seem to matter. Does R treat the character "FALSE" or "TRUE" as logical and vice versa in the appropriate context?
Ah, I can see how that would be confusing. When you use `TRUE` in `dplyr::case_when()`, that is like saying "all the rest of the time": dplyr.tidyverse.org/reference/case_when.html
@@JuliaSilge Thank you very much, I've never used case_when, but I think I will now start. This was a great video using a great data set. I'm happy to have found your channel!
@@zegpi1821 One I like is the rpart.plot shown here: www.tidymodels.org/start/tuning/#final-model Or the the ggparty plots like this: www.tidyverse.org/blog/2020/07/parsnip-0-1-2/#better-call-objects
Yes, a bagged tree uses bootstrap aggregation to make a *single prediction*. When you use bootstrap resampling with a single decision tree, you are using the resampling to estimate how well the decision tree is performing (prediction for each resample). You could use bootstrap resampling to estimate how well the predictions from a bagged tree are doing. I think the Wikipedia article on bagging is pretty good as an explanation: en.wikipedia.org/wiki/Bootstrap_aggregating
Of these, it is "Nord Polar Night Aurora". This is especially cool because you can get a Nord-Theme for other applications as well if you like the look and consistency: www.nordtheme.com/
