Closed simonpcouch closed 4 days ago
Closes #255, closes #254, closes #252, closes #233, and related to this gist.
This PR removes the add_tailor(prop) and method arguments in favor of an argument fit.workflow(calibration) giving the data to fit the calibrator on.
add_tailor(prop)
method
fit.workflow(calibration)
Benefits:
add_tailor(method)
Thanks @hfrick for this suggestion—this feels much better.
This PR should be much easier to review commit-by-commit than altogether.
TODOs from here: update tune and (possibly?) rename .should_inner_split() to something like .workflow_needs_calibration().
.should_inner_split()
.workflow_needs_calibration()
Closes #255, closes #254, closes #252, closes #233, and related to this gist.
This PR removes the
add_tailor(prop)
andmethod
arguments in favor of an argumentfit.workflow(calibration)
giving the data to fit the calibrator on.Benefits:
method
wasn't actually enough)add_tailor(method)
really wasn't truly independent of the data/resampling schemeThanks @hfrick for this suggestion—this feels much better.
This PR should be much easier to review commit-by-commit than altogether.
TODOs from here: update tune and (possibly?) rename
.should_inner_split()
to something like.workflow_needs_calibration()
.