def test_log_reg_accuracy(attribute_id, C):
data_set, labels = cd.create_data(cd.test_classes, attribute_id)
data = cd.flatten_data_set(data_set)
filename = INPUT_LOG_REG_PATTERN % (cd.attributenames[attribute_id],
str(C))
return test_classifier(data, labels, attribute_id, filename)
def train_SVM(attribute_id):
data, labels = cd.create_data(cd.train_classes, attribute_id)
train_data = cd.flatten_data_set(data)
svm = SVC(C=10., kernel='rbf', probability=True)
svm.fit(train_data, labels)
filename = OUTPUT_SVM_PATTERN % cd.attributenames[attribute_id]
bz_pickle(svm, filename)
def train_logistic_regression(attribute_id, C):
data, labels = cd.create_data(cd.train_classes, attribute_id)
train_data = cd.flatten_data_set(data)
logreg = LogisticRegression('l2', C=C, solver='saga')
logreg.fit(train_data, labels)
filename = OUTPUT_LOG_REG_PATTERN % (cd.attributenames[attribute_id],
str(C))
bz_pickle(logreg, filename)
def train_logistic_regression_CV(attribute_id):
data, labels = cd.create_data(cd.train_classes, attribute_id)
train_data = cd.flatten_data_set(data)
logreg = LogisticRegressionCV(Cs=[0.01, 0.1, 1., 10., 100.],
cv=5,
dual=False,
penalty='l2',
solver='saga',
refit=True)
logreg.fit(train_data, labels)
filename = OUTPUT_LOG_REG_CV_PATTERN % cd.attributenames[attribute_id]
bz_pickle(logreg, filename)