Catboost: CatBoostRegressor and CatBoostClassifier have not `_get_tags()`

[old-fashion-donut]

Problem: failed in sklearn RFECV.
catboost version: 0.23
Operating System: macOS 10.15.3
CPU: Intel Core i5
GPU: NaN

When try to sklearn's RFECV. got an error like below.

Example:

from catboost import CatBoostRegressor
from sklearn.model_selection import KFold
from sklearn.feature_selection import RFECV

rfecv = RFECV(estimator = CatBoostRegressor(),
              cv = KFold(5),
              scoring = 'neg_mean_squared_error')
rfecv.fit(X, y)

gives an error:

~/.pyenv/versions/3.7.7/lib/python3.7/site-packages/sklearn/feature_selection/_rfe.py in _more_tags(self)
    333 
    334     def _more_tags(self):
--> 335         estimator_tags = self.estimator._get_tags()
    336         return {'poor_score': True,
    337                 'allow_nan': estimator_tags.get('allow_nan', True)}

AttributeError: 'CatBoostRegressor' object has no attribute '_get_tags'

In Developing scikit-learn estimators,

These are annotations of estimators that allow programmatic inspection of their capabilities, such as sparse matrix support, supported output types and supported methods. The estimator tags are a dictionary returned by the method _get_tags().

sklearn RFECV checkes that CatBoost allows NaN from _get_tags method.

Can you add a _get_tags method to CatBoostRegressor and CatBoostClassifier class?

[old-fashion-donut]

I solve this problem that write this bad workaround

class CatBoostRegressor(CatBoostRegressor):
    def _get_tags(self):
        return {'allow_nan': True}

[429sugasawa]

Your answer is great.
I could confirm the operation.
CatBoost takes a lot of time, but it's a challenge.

[annaveronika]

https://github.com/catboost/catboost/pull/1133 - here is a pr that adds this method

[old-fashion-donut]

thanks @annaveronika !
and I'm sorry for that I missed #1133 pull request.