[MNT] 0.15.0 deprecation actions

[fkiraly]

Deprecation actions for the 0.15.0 release:

• changed check_is_scitype parameter msg_legacy_interface to False
• ExpandingWindowSplitter had start_with_window argument removed. From now on, initial_window=0 should be used instead of start_with_window=False. Contains bugfix [BUG] fix start_with_window deprecation in ExpandingWindowSplitter #3953 as the deprecation logic was incomplete.
• Test test_methods_do_not_change_state renamed to test_non_state_changing_method_contract
• row transformers have been removed, replaced by base class vectorization/broadcasting
• changed default for default_fc_parameters in tsfresh transformers to follow the default of tsfresh, which currently is comprehensive
• removed module transformations.series.compose. All contents were moved to transformations.compose.

Delayed to 0.16.0:

• base class of ProbabilityThresholdEarlyClassifier change to BaseEarlyClassifier. Unclear what to do here, deprecation action specified breaks the interface compliance tests. Attempt to fix here [BUG] fix faulty deprecation for ProbabilityThresholdEarlyClassifier base class change #3962 but fails fit_idempotent test (unclear why, will drop for this release).
• change of forecaster cutoff attribute to pd.Index type

[review-notebook-app]

Check out this pull request on 

See visual diffs & provide feedback on Jupyter Notebooks.

---

Powered by ReviewNB

[miraep8]

Was the depreciation also meant to extend to start_with_window in SlidingWindowSplitter as well? (Or should we perhaps introduce a depreciation for this?). Just thinking that that splitter also has an initial window/could make sense to expand up to window_length then begin sliding?

[fkiraly]

Was the depreciation also meant to extend to start_with_window in SlidingWindowSplitter as well?

It wasn't, since we have not raised a deprecation warning for it.

I have also seen the same thing while going through the code, this could have been done at the same time (but then again, the ExpandingWindowSplitter was not deprecated correctly).

Probably worth opening an issue on this and discussing whether the parameter is also redundant in SlidingWindowSplitter. I think it is, by virtue of the same argument (if False just set the other parameter to 0).

[miraep8]

Thank goodness we finally got those OptionalPassthrough imports in the correct order.

[fkiraly]

Thank goodness we got those OptionalPassthrough imports in the correct order.

Well, the location moved, and with it the import's position in the lexicographical order.