More survival augment testing #178
Conversation
| Warning in `show_best()`: | ||
| 4 evaluation times are available; the first will be used (i.e. `eval_time = 10`). |
There was a problem hiding this comment.
Does this seem right? Should we modify augment.tune_results() such that this doesn't happen. There are currently no way of avoiding this as augment.tune_results() doesn't take the arguments it need
There was a problem hiding this comment.
That's tidymodels/tune#810
We want augment() to treat eval_time like it treats metric: if you don't specify which parameters you want to use, it will use the first metric and pass that to select_best(). Analogous, it should pick the first eval_time and pass that. Both parts documented, of course.
There was a problem hiding this comment.
got it! augment.tune_results() doesn't have a metric argument, instead it has a parameters argument. These are now tested with fa2099e (#178)
There was a problem hiding this comment.
Added suggested changes for tidymodels/tune#832 as an alternative to tidymodels/tune#825
|
Looks like the failures on CI are related to the tune exit handlers rather than anything related to survival analysis. |
To close #132