def _calibrate_model(self, model, method: str, X_valid: np.array, y_valid: np.array):
"""
Note that we use _CalibratedClassifier to better deal with the label encoding.
Arguments:
model {BaseEstimator} -- sklearn model model
method {str} -- [description]
X_valid {np.array} -- Feature matrix
y_valid {np.array} -- Label vector
Returns:
[_CalibratedClassifier] -- calibrated classifier
"""
if method == 'isotonic':
calibrated = _CalibratedClassifier(model, method='isotonic', classes=self.classes)
calibrated.fit(X_valid, y_valid)
elif method == 'sigmoid':
calibrated = _CalibratedClassifier(model, method='sigmoid', classes=self.classes)
calibrated.fit(X_valid, y_valid)
elif method == 'none':
calibrated = model
else:
calibrated = _CalibratedClassifier(model, method='sigmoid', classes=self.classes)
calibrated.fit(X_valid, y_valid)
return calibrated
def _calibrate_model(self, model, method: str, X_valid: np.array,
y_valid: np.array):
"""
Note that we use _CalibratedClassifier to better deal with the label encoding.
Arguments:
model (BaseEstimator) -- sklearn model model
method (str) -- Calibration method "isotonic" or "sigmoid"
X_valid (np.array) -- Feature matrix
y_valid (np.array) -- Label vector
Returns:
_CalibratedClassifier -- calibrated classifier
"""
if method == "isotonic":
calibrated = _CalibratedClassifier(model,
method="isotonic",
classes=self.classes)
calibrated.fit(X_valid, y_valid)
elif method == "sigmoid":
calibrated = _CalibratedClassifier(model,
method="sigmoid",
classes=self.classes)
calibrated.fit(X_valid, y_valid)
elif method == "none":
calibrated = model
else:
calibrated = _CalibratedClassifier(model,
method="sigmoid",
classes=self.classes)
calibrated.fit(X_valid, y_valid)
return calibrated
def test_calibration_zero_probability():
# Test an edge case where _CalibratedClassifier avoids numerical errors
# in the multiclass normalization step if all the calibrators output
# are zero all at once for a given sample and instead fallback to uniform
# probabilities.
class ZeroCalibrator:
# This function is called from _CalibratedClassifier.predict_proba.
def predict(self, X):
return np.zeros(X.shape[0])
X, y = make_blobs(n_samples=50,
n_features=10,
random_state=7,
centers=10,
cluster_std=15.0)
clf = DummyClassifier().fit(X, y)
calibrator = ZeroCalibrator()
cal_clf = _CalibratedClassifier(base_estimator=clf,
calibrators=[calibrator],
classes=clf.classes_)
probas = cal_clf.predict_proba(X)
# Check that all probabilities are uniformly 1. / clf.n_classes_
assert_allclose(probas, 1.0 / clf.n_classes_)
def _calibrate_model(self, model, method: str, X_valid: np.array,
y_valid: np.array):
"""
Note that we use _CalibratedClassifier to better deal with the label encoding.
Arguments:
model {BaseEstimator} -- sklearn model model
method {str} -- [description]
X_valid {np.array} -- Feature matrix
y_valid {np.array} -- Label vector
Returns:
[_CalibratedClassifier] -- calibrated classifier
"""
if method == "isotonic":
calibrated = _CalibratedClassifier(model,
method="isotonic",
classes=self.classes)
calibrated.fit(X_valid, y_valid)
elif method == "sigmoid":
calibrated = _CalibratedClassifier(model,
method="sigmoid",
classes=self.classes)
calibrated.fit(X_valid, y_valid)
elif method == "none":
calibrated = model
elif method == "spline":
warnings.warn("spline is deprecated option!", DeprecationWarning)
calibrated = SplineCalibratedClassifierCV(model, cv="prefit")
calibrated.fit(X_valid, y_valid)
else:
calibrated = _CalibratedClassifier(model,
method="sigmoid",
classes=self.classes)
calibrated.fit(X_valid, y_valid)
return calibrated