def test__prediction_to_row(self):
repeat_nr = 0
fold_nr = 0
clf = sklearn.pipeline.Pipeline(
steps=[('Imputer', Imputer(strategy='mean')
), ('VarianceThreshold', VarianceThreshold(
threshold=0.05)), ('Estimator', GaussianNB())])
task = openml.tasks.get_task(20)
train, test = task.get_train_test_split_indices(repeat_nr, fold_nr)
X, y = task.get_X_and_y()
clf.fit(X[train], y[train])
test_X = X[test]
test_y = y[test]
probaY = clf.predict_proba(test_X)
predY = clf.predict(test_X)
sample_nr = 0 # default for this task
for idx in range(0, len(test_X)):
arff_line = _prediction_to_row(repeat_nr, fold_nr, sample_nr, idx,
task.class_labels[test_y[idx]],
predY[idx], probaY[idx],
task.class_labels, clf.classes_)
self.assertIsInstance(arff_line, list)
self.assertEqual(len(arff_line), 6 + len(task.class_labels))
self.assertEqual(arff_line[0], repeat_nr)
self.assertEqual(arff_line[1], fold_nr)
self.assertEqual(arff_line[2], sample_nr)
self.assertEqual(arff_line[3], idx)
sum = 0.0
for att_idx in range(4, 4 + len(task.class_labels)):
self.assertIsInstance(arff_line[att_idx], float)
self.assertGreaterEqual(arff_line[att_idx], 0.0)
self.assertLessEqual(arff_line[att_idx], 1.0)
sum += arff_line[att_idx]
self.assertAlmostEqual(sum, 1.0)
self.assertIn(arff_line[-1], task.class_labels)
self.assertIn(arff_line[-2], task.class_labels)
pass
def test__prediction_to_row(self):
repeat_nr = 0
fold_nr = 0
clf = sklearn.pipeline.Pipeline(steps=[('Imputer', Imputer(strategy='mean')),
('VarianceThreshold', VarianceThreshold(threshold=0.05)),
('Estimator', GaussianNB())])
task = openml.tasks.get_task(20)
train, test = task.get_train_test_split_indices(repeat_nr, fold_nr)
X, y = task.get_X_and_y()
clf.fit(X[train], y[train])
test_X = X[test]
test_y = y[test]
probaY = clf.predict_proba(test_X)
predY = clf.predict(test_X)
for idx in range(0, len(test_X)):
arff_line = _prediction_to_row(repeat_nr, fold_nr, idx,
task.class_labels[test_y[idx]],
predY[idx], probaY[idx], task.class_labels, clf.classes_)
self.assertIsInstance(arff_line, list)
self.assertEqual(len(arff_line), 5 + len(task.class_labels))
self.assertEqual(arff_line[0], repeat_nr)
self.assertEqual(arff_line[1], fold_nr)
self.assertEqual(arff_line[2], idx)
sum = 0.0
for att_idx in range(3, 3 + len(task.class_labels)):
self.assertIsInstance(arff_line[att_idx], float)
self.assertGreaterEqual(arff_line[att_idx], 0.0)
self.assertLessEqual(arff_line[att_idx], 1.0)
sum += arff_line[att_idx]
self.assertAlmostEqual(sum, 1.0)
self.assertIn(arff_line[-1], task.class_labels)
self.assertIn(arff_line[-2], task.class_labels)
pass