ordinal regression model type & polr engine #8
Conversation
ordinal regression model type & polr engine
There was a problem hiding this comment.
Overall it looks good. @ me if you have questions or need help with anything.
A few random things unrelated to the PR:
- The dev version of yardstick has the function
ranked_prob_scorethat is specific to ordinal models. - Once this package is on CRAN, I’ll make “engine” documentation files for each model/engine combo here. Here’s an example. I’ll add you to review so you can check them and make edits as you see fit.
- Before sending to CRAN, check out Davis's extrachecks repo.
| functionality. See below for examples of fitting ordinal regression models | ||
| with {censored}. | ||
| } | ||
| \examples{ |
There was a problem hiding this comment.
I added some notes about "noSuggests" to the test files. That comes up here too.
There are data and packages that are used here and CRAN will complain since the packages may not be installed by the user.
There are two ways to deal with this.
- Wrap the whole thing in a loop that uses
if (rlang::is_installed(c("MASS", "ordinalForest"))). That's easy and here are examples - More crafty is to prevent CRAN from seeing the examples. This might bite you on the initial submission but here is how to do that.
| Version: 0.0.0.9000 | ||
| Authors@R: c( | ||
| person("Max", "Kuhn", , "[email protected]", role = c("aut", "cre"), | ||
| comment = c(ORCID = "0000-0003-2402-136X")), | ||
| person("Jason Cory", "Brunson", , "[email protected]", role = c("aut"), | ||
| comment = c(ORCID = "0000-0003-3126-9494")), | ||
| person("Posit Software PBC", role = "cph") |
There was a problem hiding this comment.
You can remove this since it is yours.
I'd also run usethis::use_tidy_description() to reorganize this file.
There was a problem hiding this comment.
@topepo i'm not sure what you mean by "remove this".
Also, should i mark myself "cre" for the time being or wait until someone is ready to submit to CRAN?
| #' functionality. See below for examples of fitting ordinal regression models | ||
| #' with {censored}. | ||
| #' | ||
| #' @examples |
There was a problem hiding this comment.
Same things related to "examplesif" and is_installed() apply here.
| #' {ordered} provides engines for ordinal regression models for the {parsnip} | ||
| #' package. The models may have cumulative, sequential, or adjacent-category | ||
| #' structure, and in future these may be disaggregated into separate model | ||
| #' types. A vignette will provide thorough illustrations of {ordered} |
There was a problem hiding this comment.
Yet! Should we require a vignette prior to initial submission?
| type = "class", | ||
| value = list( | ||
| pre = NULL, | ||
| post = NULL, |
There was a problem hiding this comment.
Do we need to reconvert the predictions to be ordered factors?
This branch might not be in final form, but should i synchronize the fork and submit a PR (and into what branch of |
This PR continues #6; see there for details.