def test_multiclass_prediction(self, predict_mock):
classes = [['a', 'b', 'c']]
predicted_probabilities = [[0, 0, 0.99], [0, 0.99, 0], [0.99, 0, 0],
[0, 0.99, 0], [0, 0, 0.99]]
predicted_indexes = [2, 1, 0, 1, 2]
expected_result = ['c', 'b', 'a', 'b', 'c']
predict_mock.return_value = np.array(predicted_probabilities)
classifier = AutoMLClassifier(
time_left_for_this_task=1,
per_run_time_limit=1,
backend=None,
)
classifier._classes = [np.array(classes)]
classifier._n_outputs = 1
classifier._n_classes = np.array([3])
actual_result = classifier.predict([None] * len(predicted_indexes))
np.testing.assert_array_equal(expected_result, actual_result)
def test_multiclass_prediction(self, predict_mock):
predicted_probabilities = [[0, 0, 0.99], [0, 0.99, 0], [0.99, 0, 0],
[0, 0.99, 0], [0, 0, 0.99]]
predicted_indexes = [2, 1, 0, 1, 2]
expected_result = ['c', 'b', 'a', 'b', 'c']
predict_mock.return_value = np.array(predicted_probabilities)
classifier = AutoMLClassifier(
time_left_for_this_task=1,
per_run_time_limit=1,
backend=None,
)
classifier.InputValidator.validate_target(
pd.DataFrame(expected_result, dtype='category'),
is_classification=True,
)
actual_result = classifier.predict([None] * len(predicted_indexes))
np.testing.assert_array_equal(expected_result, actual_result)
def test_multiclass_prediction(self, predict_mock):
classes = [['a', 'b', 'c']]
predicted_probabilities = [[0, 0, 0.99], [0, 0.99, 0], [0.99, 0, 0],
[0, 0.99, 0], [0, 0, 0.99]]
predicted_indexes = [2, 1, 0, 1, 2]
expected_result = ['c', 'b', 'a', 'b', 'c']
predict_mock.return_value = np.array(predicted_probabilities)
classifier = AutoMLClassifier(
time_left_for_this_task=1,
per_run_time_limit=1,
backend=None,
)
classifier._classes = [np.array(classes)]
classifier._n_outputs = 1
classifier._n_classes = np.array([3])
actual_result = classifier.predict([None] * len(predicted_indexes))
np.testing.assert_array_equal(expected_result, actual_result)
def test_multilabel_prediction(self, predict_mock):
classes = [[1, 2], [13, 17]]
predicted_probabilities = [[0.99, 0], [0.99, 0], [0, 0.99],
[0.99, 0.99], [0.99, 0.99]]
predicted_indexes = np.array([[1, 0], [1, 0], [0, 1], [1, 1], [1, 1]])
expected_result = np.array([[2, 13], [2, 13], [1, 17], [2, 17],
[2, 17]])
predict_mock.return_value = np.array(predicted_probabilities)
classifier = AutoMLClassifier(
time_left_for_this_task=1,
per_run_time_limit=1,
backend=None,
)
classifier._classes = list(map(np.array, classes))
classifier._n_outputs = 2
classifier._n_classes = np.array([3, 2])
actual_result = classifier.predict([None] * len(predicted_indexes))
np.testing.assert_array_equal(expected_result, actual_result)
def test_multilabel_prediction(predict_mock, backend, dask_client):
predicted_probabilities = [[0.99, 0], [0.99, 0], [0, 0.99], [0.99, 0.99],
[0.99, 0.99]]
predicted_indexes = np.array([[1, 0], [1, 0], [0, 1], [1, 1], [1, 1]])
expected_result = np.array([[2, 13], [2, 13], [1, 17], [2, 17], [2, 17]])
predict_mock.return_value = np.array(predicted_probabilities)
classifier = AutoMLClassifier(
time_left_for_this_task=1,
per_run_time_limit=1,
backend=backend,
dask_client=dask_client,
)
classifier.InputValidator.validate_target(
pd.DataFrame(expected_result, dtype='int64'),
is_classification=True,
)
actual_result = classifier.predict([None] * len(predicted_indexes))
np.testing.assert_array_equal(expected_result, actual_result)
def test_multilabel_prediction(predict_mock, dask_client):
predicted_probabilities = [[0.99, 0], [0.99, 0], [0, 0.99], [0.99, 0.99],
[0.99, 0.99]]
predicted_indexes = np.array([[1, 0], [1, 0], [0, 1], [1, 1], [1, 1]])
predict_mock.return_value = np.array(predicted_probabilities)
classifier = AutoMLClassifier(
time_left_for_this_task=1,
per_run_time_limit=1,
dask_client=dask_client,
)
classifier.InputValidator = InputValidator(is_classification=True)
classifier.InputValidator.target_validator.fit(
pd.DataFrame(predicted_indexes, dtype='int64'), )
classifier.InputValidator._is_fitted = True
assert classifier.InputValidator.target_validator.type_of_target == 'multilabel-indicator'
actual_result = classifier.predict([None] * len(predicted_indexes))
np.testing.assert_array_equal(predicted_indexes, actual_result)
def test_multilabel_prediction(self, predict_mock):
classes = [[1, 2], [13, 17]]
predicted_probabilities = [[0.99, 0],
[0.99, 0],
[0, 0.99],
[0.99, 0.99],
[0.99, 0.99]]
predicted_indexes = np.array([[1, 0], [1, 0], [0, 1], [1, 1], [1, 1]])
expected_result = np.array([[2, 13], [2, 13], [1, 17], [2, 17], [2, 17]])
predict_mock.return_value = np.array(predicted_probabilities)
classifier = AutoMLClassifier(
time_left_for_this_task=1,
per_run_time_limit=1,
backend=None,
)
classifier._classes = list(map(np.array, classes))
classifier._n_outputs = 2
classifier._n_classes = np.array([3, 2])
actual_result = classifier.predict([None] * len(predicted_indexes))
np.testing.assert_array_equal(expected_result, actual_result)