explain_prediction with OneVsRestClassifier and SVC

[oterrier]

Hi Thx for this great package It looks like there is an error when you use OneVsRestClassifier with SVCClassifier

I have this error

    if clf.kernel != 'linear':
AttributeError: 'OneVsRestClassifier' object has no attribute 'kernel'

Here is the little code snippet to reproduce the issue:

from sklearn.datasets import make_multilabel_classification, fetch_20newsgroups
from sklearn.feature_extraction.text import TfidfVectorizer
from sklearn.multiclass import OneVsRestClassifier
from sklearn.svm import SVC

data_train = fetch_20newsgroups(subset='train', shuffle=True, random_state=42)
vectorizer = TfidfVectorizer(stop_words='english')
X_train = vectorizer.fit_transform(data_train.data[0:100])
y_train = data_train.target[0:100]
classif = OneVsRestClassifier(SVC(kernel='linear'))
classif.fit(X_train, y_train)
expl = eli5.explain_prediction(classif, data_train.data[0], vec=vectorizer)

Thx a lot again

Olivier

[oterrier]

Hi If you change SVC by LinearSVC in the previous code snippet, it works ... Any reason SVC is not supported the same way?

Best regards Olivier