Simon P. Couch

Seeing the following deprecation condition while testing tailor: ``` transpose() is superseded. Please use `list_transpose()` instead. ``` Traceback: ``` 8. ├─generics::fit(tlr, d_calibration, outcome = y, estimate = y_pred) 9. ├─tailor:::fit.tailor(tlr,...

inline source code for overfitting plots

1

A little bit of developer complexity for the sake of participant simplicity. Given that we're already teaching a binary classification problem, I thought it'd be nice if the colors matched...

workers error with custom metrics when using socket clusters

1

A revival of #479, reprex whittled down from tidymodels/finetune#116, will also fix tidymodels/yardstick#514 when addressed. When using socket cluster parallelism (notably, this not an issue with forking), workers can't find...

Closes #255. Opted to just check for non-`NULL` arguments rather than missingness to allow for passing arguments programmatically more easily. "Does not require training" also doesn't feel quite informative enough,...

Same goes for `add_tailor(method)`: if the workflow doesn't require an inner split but arguments are passed related to it, we ought to error in `add_tailor()`.

`make_inner_split()` errors for some `add_tailor(method)` values

1

When the rsample function called internally by `inner_split()` methods doesn't have a `prop` argument, that function errors. e.g. with `method = "boot_split"`: ``` r library(tidymodels) library(tailor) tlr % adjust_numeric_calibration() %>%...

Currently just reprints the tailor, but may be worth displaying the proportion and method when printing the workflow.

address `fit()` slowdown with sparse tibble and formula preprocessor

4

Related to #239—just a place to keep notes on the thought process for supporting sparse tibbles with formula preprocessors. In #245, we see: ``` r library(tidymodels) sparse_hotel_rates

look into reintroducing support for `margins`

1

Related to #1200--the package is now back on CRAN.