def fit_and_validate(self):
'''
Returns training & testing log loss
'''
X_train, X_test = self._pick_examples()
# shorthand
Y_train = self.Y_train
Y_test = self.Y_test
self.clf.fit(X_train, Y_train)
if self.scoring == 'accuracy':
a_train = metrics.accuracy_score(Y_train,
self.clf.predict(X_train))
a_test = np.array(
[]) if isEmpty(Y_test) else metrics.accuracy_score(
Y_test, self.clf.predict(X_test))
return a_train, a_test
else: #log_loss
# get probabilities
self._proba_train = self.clf.predict_proba(X_train)
self._proba_test = self.clf.predict_proba(X_test)
return metrics.log_loss(Y_train, self.proba_train), np.array(
[]) if isEmpty(Y_test) else metrics.log_loss(
Y_test, self.proba_test)
def _pick_examples(self):
'''
If we have scaled examples - pick them, else pick X_train, X_test
'''
return (self.X_train, self.X_test) \
if isEmpty(self.X_train_scaled) or isEmpty(self.X_test_scaled) \
else (self.X_train_scaled, self.X_test_scaled)
def _pick_examples(self):
'''
If we have scaled examples - pick them, else pick X_train, X_test
'''
return (self.X_train, self.X_test) \
if isEmpty(self.X_train_scaled) or isEmpty(self.X_test_scaled) \
else (self.X_train_scaled, self.X_test_scaled)
def fit_and_validate(self):
'''
Returns training & testing log loss
'''
X_train, X_test = self._pick_examples()
# shorthand
Y_train = self.Y_train
Y_test = self.Y_test
self.clf.fit(X_train, Y_train)
if self.scoring == 'accuracy':
a_train = metrics.accuracy_score(Y_train, self.clf.predict(X_train))
a_test = np.array([]) if isEmpty(Y_test) else metrics.accuracy_score(Y_test, self.clf.predict(X_test))
return a_train, a_test
else: #log_loss
# get probabilities
self._proba_train = self.clf.predict_proba(X_train)
self._proba_test = self.clf.predict_proba(X_test)
return metrics.log_loss(Y_train, self.proba_train), np.array([]) if isEmpty(Y_test) else metrics.log_loss(Y_test, self.proba_test)
def fit_and_validate(self):
'''
Returns training & testing log loss
'''
X_train, X_test = self._pick_examples()
# shorthand
Y_train = self.Y_train
Y_test = self.Y_test
self.clf.fit(X_train, Y_train)
# get probabilities
self._proba_train = self.clf.predict_proba(X_train)
self._proba_test = self.clf.predict_proba(X_test)
return metrics.log_loss(Y_train, self.proba_train), np.array([]) if isEmpty(Y_test) else metrics.log_loss(Y_test, self.proba_test)
def fit_and_validate(self):
'''
Returns training & testing log loss
'''
X_train, X_test = self._pick_examples()
# shorthand
Y_train = self.Y_train
Y_test = self.Y_test
self.clf.fit(X_train, Y_train)
# get probabilities
self._proba_train = self.clf.predict_proba(X_train)
self._proba_test = self.clf.predict_proba(X_test)
return metrics.log_loss(Y_train, self.proba_train), np.array(
[]) if isEmpty(Y_test) else metrics.log_loss(
Y_test, self.proba_test)
