[DOC] Inconsistency in documented `TimeSeriesForestClassifier` locations, sktime 0.25.0 docs about sktime HOT 6 CLOSED

Fixed, here: #5741

What is odd, the module was never called classifiers afaik, this must have been a mistake in the original docstring.

from sktime.

Thanks fo reporting!

• the module starting with underscore is private, so should not be directly accessed
• the note sems outdated and ought to be fixed, though not sure what exactly it should change to

from sktime.

• the module starting with underscore is private, so should not be directly accessed

In this case, the underscore form should not appear in the all_estimators() output, especially when the other (recommended) version does not appear in that output? Otherwise, users might just use the form of estimator they see output in this "encyclopaedia" of methods?

from sktime.

In this case, the underscore form should not appear in the all_estimators() output, especially when the other (recommended) version does not appear in that output?

Interesting, where does it appear?

If you convert a class to str, afaik it will always appear with its full path.

from sktime.

You can see what I describe after

from sktime.registry import all_estimators
all_estimators('classifier')

Starting from
"('Arsenal', sktime.classification.kernel_based._arsenal.Arsenal)"
there are many paths with the _ shown.

from sktime.

yes, I see - that's because the second element in the tuple is a class. When converted to string, the full path is always shown. Not sure how to change that.

str(all_estimators('classifier')[0][1]) gives <class 'sktime.classification.kernel_based._arsenal.Arsenal'>

Suggestions welcome.

from sktime.