TypeError: predict_proba not available for this estimator

TypeError: predict_proba unavailable

$ python -c "from sklearn.svm import SVC; SVC().predict_proba([[0,1]])"
Traceback (most recent call last):
  File "<string>", line 1, in <module>
AttributeError: 'SVC' object has no attribute 'predict_proba'

Why this happens

• Estimator doesn’t implement probability estimates by default.

Fix

• Use probability=True for SVC, or choose an estimator that supports it.

Wrong code

from sklearn.svm import SVC
SVC().predict_proba([[0,1]])

Fixed code

from sklearn.svm import SVC
clf = SVC(probability=True)
clf.fit([[0,1],[1,0],[0.5,0.5]],[0,1,0])
print(clf.predict_proba([[0.2,0.8]]))