Scikit learn: which regressors natively support multi-target regression?

Question

The docs on sklearn.multioutput.MultiOutputRegressor state that it implements a strategy for extending regressors that do not natively support multi-target regression.

I'm interested to know: which ones do natively support multi-target regression ? (and/or how can I find out?)

based on this article, I assume:

ExtraTreesRegressor
KNeighborsRegressor
LinearRegression
RidgeCV

All support this? (unless I'm getting confused and a 'multi-output estimator' is not the same thing as 'a regressor that natively supports multi-target regression'; I'm new to this!)

Answer 1

In scikit-learn, there is a MultiOutputMixin class. So, you could check if the estimator inherits from that one.

Alternatively, you could use estimator tags and check if {'multioutput': True} is included in the tags of the estimator.