SparseCoder does not follow the scikit-learn API and will break in 0.24 #16336
Description
The SparseCoder does not follow the scikit-learn API.
d = np.array([[1, 2, 3], [1, 2, 3]])
sc = SparseCoder(dictionary=d)
sc.get_params()
/home/glemaitre/Documents/packages/scikit-learn/sklearn/base.py:204:
FutureWarning: From version 0.24, get_params will raise an AttributeError
if a parameter cannot be retrieved as an instance attribute.
Previously it would return None. FutureWarning)The reason is that we initialize self.components_ = dictionary meaning that self.dictionary does not exist which raises the warning and will error in 0.24.
Basically, set_params will not work as well then.
So we should probably solve this issue.