test fit_best.workflow_set() with survival
#200
Conversation
| ) | ||
|
|
||
| expect_silent(dynamic_res <- fit_best(grid_dynamic_res)) | ||
| dynamic_res <- fit_best(grid_dynamic_res, eval_time = 1) |
There was a problem hiding this comment.
I added these two lines to mirror the existing fit_best() tests but am a bit unsure of the intent.
There was a problem hiding this comment.
The expect_silent() is there because we want this function to make a choice for metric and eval_time silently, unlike most other functions where it warns you about that.
The second one I'm not entirely sure about. Ideally, we'd test that it does indeed pick the best model for a specific eval time (but there is nothing currently confirming that) so it's just making sure that it doesn't error 🤔
| ) | ||
|
|
||
| expect_silent(dynamic_res <- fit_best(grid_dynamic_res)) | ||
| dynamic_res <- fit_best(grid_dynamic_res, eval_time = 1) |
There was a problem hiding this comment.
The expect_silent() is there because we want this function to make a choice for metric and eval_time silently, unlike most other functions where it warns you about that.
The second one I'm not entirely sure about. Ideally, we'd test that it does indeed pick the best model for a specific eval time (but there is nothing currently confirming that) so it's just making sure that it doesn't error 🤔
|
Just started rerunning checks with the most recent finetune commit on that branch. :) |
Mirrors the tests for
fit_best.tune_results()as closely as possible, switching out model + preprocessor with a workflowset containing the same!Tests for tidymodels/workflowsets#144.