import numpy as np
from sklearn import grid_search
from sklearn.linear_model import SGDClassifier
from sklearn.pipeline import make_pipeline
from numpy.random import random, random_integers

X = random([300,1000])
y = random_integers(0, 1, [300, ])
param_grid = {
    'loss': ['log'],
    'penalty': ['elasticnet'],
    'alpha': [10 ** x for x in range(-6, 1)],
    'l1_ratio': [0, 0.05, 0.1, 0.2, 0.5, 0.8, 0.9, 0.95, 1],
}
clf = SGDClassifier(random_state=0, class_weight='balanced')

clf_grid = grid_search.GridSearchCV(estimator=clf, param_grid=param_grid,
                                    n_jobs=-1, scoring='roc_auc')

clf_grid.fit(X=X, y=y)
clf_grid.predict_proba(X)

  File "<ipython-input-1-c4ceb6ddb8c1>", line 1, in <module>
    runfile('/home/yichuanliu/Dropbox/Python/Cognoma/test.py', wdir='/home/yichuanliu/Dropbox/Python/Cognoma')

  File "/home/yichuanliu/Programs/anaconda3/lib/python3.5/site-packages/spyderlib/widgets/externalshell/sitecustomize.py", line 699, in runfile
    execfile(filename, namespace)

  File "/home/yichuanliu/Programs/anaconda3/lib/python3.5/site-packages/spyderlib/widgets/externalshell/sitecustomize.py", line 88, in execfile
    exec(compile(open(filename, 'rb').read(), filename, 'exec'), namespace)

  File "/home/yichuanliu/Dropbox/Python/Cognoma/test.py", line 28, in <module>
    clf_grid.predict_proba(X)

  File "/home/yichuanliu/Programs/anaconda3/lib/python3.5/site-packages/sklearn/utils/metaestimators.py", line 35, in __get__
    self.get_attribute(obj)

  File "/home/yichuanliu/Programs/anaconda3/lib/python3.5/site-packages/sklearn/linear_model/stochastic_gradient.py", line 756, in predict_proba
    self._check_proba()

  File "/home/yichuanliu/Programs/anaconda3/lib/python3.5/site-packages/sklearn/linear_model/stochastic_gradient.py", line 721, in _check_proba
    " loss=%r" % self.loss)

AttributeError: probability estimates are not available for loss='hinge'

clf_grid.best_estimator_
Out[4]: 
SGDClassifier(alpha=0.1, average=False, class_weight='balanced', epsilon=0.1,
       eta0=0.0, fit_intercept=True, l1_ratio=0.1, learning_rate='optimal',
       loss='log', n_iter=5, n_jobs=1, penalty='elasticnet', power_t=0.5,
       random_state=0, shuffle=True, verbose=0, warm_start=False)

param_grid = {
    'alpha': [10 ** x for x in range(-6, 1)],
    'l1_ratio': [0, 0.05, 0.1, 0.2, 0.5, 0.8, 0.9, 0.95, 1],
}
clf = SGDClassifier(random_state=0, class_weight='balanced',
                    loss='log', penalty='elasticnet')

clf_grid = grid_search.GridSearchCV(estimator=clf, param_grid=param_grid,
                                    n_jobs=-1, scoring='roc_auc')
clf_grid.fit(X=X, y=y)
clf_grid.predict_proba(X)

class SGDClassifier(BaseSGDClassifier, _LearntSelectorMixin):
...
    @property
    def predict_proba(self):
        """Probability estimates.
        ...
        """
        self._check_proba()
        return self._predict_proba

class SGDClassifier(BaseSGDClassifier, _LearntSelectorMixin):
...
    def predict_proba(self, X):
        """Probability estimates.
        ...
        """
        self._check_proba()
        return self._predict_proba(X)