This check fails when transform in invoked by ember on this line (in ember/features.py, line 192):
entry_name_hashed = FeatureHasher(50, input_type="string").transform([raw_obj['entry']]).toarray()[0]
because [raw_obj['entry']] is a list of strings, not a list of list of strings.
This pull request changes this call to transform by wrapping everything in a list.
I am not sure of the soundness of my fix, so I encourage the reviewer to have a deeper look.
A recent version of scikit-learn added a check to the input data of the function "transform" in FeatureHasher. The details are in this pull request: https://github.com/scikit-learn/scikit-learn/pull/25094.
This check fails when transform in invoked by ember on this line (in ember/features.py, line 192):
entry_name_hashed = FeatureHasher(50, input_type="string").transform([raw_obj['entry']]).toarray()[0]
because[raw_obj['entry']]
is a list of strings, not a list of list of strings.This pull request changes this call to transform by wrapping everything in a list. I am not sure of the soundness of my fix, so I encourage the reviewer to have a deeper look.