def test_robust_label_encoder_unsorted_labels_warning(labels):
enc = RobustLabelEncoder(labels=labels)
with pytest.warns(UserWarning):
enc.fit([labels[2], labels[0]])
assert_array_equal(list(enc.classes_), sorted(labels))
assert_array_equal(enc.transform([labels[1], labels[2], "173"]), [2, 1, 3])
def test_robust_label_encoder():
enc = RobustLabelEncoder()
enc.fit(X[:, 0])
assert_array_equal(enc.classes_, ["apple", "banana", "hot dog"])
assert_array_equal(enc.transform([]), [])
assert_array_equal(enc.transform(["hot dog", "banana", "hot dog"]), [2, 1, 2])
assert_array_equal(enc.transform(["hot dog", "llama"]), [2, 3])
assert_array_equal(enc.inverse_transform([0, 2]), ["apple", "hot dog"])
assert_array_equal(enc.inverse_transform([0, 10]), ["apple", "<unseen_label>"])
assert_array_equal(enc.fit_transform(X[:, 0]), [2, 2, 0, 2, 2, 1])
def test_robust_label_encoder_fill_label_value():
y = np.array([1, 1, 0, 1, 1])
enc = RobustLabelEncoder(labels=[1], fill_label_value=0)
enc.fit(y)
y_transform = enc.transform(y)
assert_array_equal(y_transform, [0, 0, 1, 0, 0])
assert_array_equal(enc.inverse_transform(y_transform), y)
# Test that fit_transform has the same behavior
enc = RobustLabelEncoder(labels=[1], fill_label_value=0)
y_transform = enc.fit_transform(y)
assert_array_equal(y_transform, [0, 0, 1, 0, 0])
assert_array_equal(enc.inverse_transform(y_transform), y)
def build_label_transform():
"""Returns the model definition representing feature processing."""
return RobustLabelEncoder(labels=[
'LAYING', 'SITTING', 'STANDING', 'WALKING', 'WALKING_DOWNSTAIRS',
'WALKING_UPSTAIRS'
])
def test_robust_label_encoder_sorted_labels(labels):
enc = RobustLabelEncoder(labels=labels)
enc.fit([labels[1], labels[0]])
assert_array_equal(list(enc.classes_), labels)
assert_array_equal(enc.get_classes(), labels)
assert_array_equal(enc.transform([labels[2], labels[1], "173"]), [2, 1, 3])
# Test that fit_transform has the same behavior
enc = RobustLabelEncoder(labels=labels)
y_transformed = enc.fit_transform([labels[2], labels[1], "173"])
assert_array_equal(list(enc.classes_), labels)
assert_array_equal(y_transformed, [2, 1, 3])
def test_robust_label_encoder_inverse_transform_dtype():
enc = RobustLabelEncoder()
enc.fit(X[:, 0])
np.testing.assert_array_equal(enc.inverse_transform(["1.0", "2.0"]), ["banana", "hot dog"])
with pytest.raises(ValueError):
enc.inverse_transform(["0.", "2.b"])
def test_robust_label_encoder_unsorted_labels_warning(labels):
enc = RobustLabelEncoder(labels=labels)
with pytest.warns(UserWarning):
enc.fit([labels[2], labels[0]])
assert_array_equal(list(enc.classes_), sorted(labels))
assert_array_equal(enc.transform([labels[1], labels[2], "173"]), [2, 1, 3])
# Test that fit_transform has the same behavior
enc = RobustLabelEncoder(labels=labels)
with pytest.warns(UserWarning):
y_transformed = enc.fit_transform([labels[1], labels[2], "173"])
assert_array_equal(list(enc.classes_), sorted(labels))
assert_array_equal(y_transformed, [2, 1, 3])
def test_inverse_label_transformer():
st_helper = SklearnTestHelper()
rle = RobustLabelEncoder()
# Binary Classification
data = np.random.random_sample((10, )).astype(np.float32)
dshape = (relay.Any(), )
st_helper.compile(rle, dshape, "float32", "inverse_transform")
python_out = (data > 0.5).astype(int)
tvm_out = st_helper.run(data)
tvm.testing.assert_allclose(python_out, tvm_out, rtol=1e-5, atol=1e-5)
# Multiclass Classification
data = np.random.random_sample((10, 5)).astype(np.float32)
dshape = (relay.Any(), 5)
st_helper.compile(rle, dshape, "float32", "inverse_transform")
python_out = np.argmax(data, axis=1)
tvm_out = st_helper.run(data)
tvm.testing.assert_allclose(python_out, tvm_out, rtol=1e-5, atol=1e-5)
def build_label_transform():
"""Returns the model definition representing feature processing."""
return RobustLabelEncoder(labels=["1", "2", "3", "4", "5"])
def test_robust_label_encoder_unsorted_labels_warning(labels):
enc = RobustLabelEncoder(labels=labels)
with pytest.warns(UserWarning):
enc.fit([labels[2], labels[0]])
assert_array_equal(list(enc.classes_), sorted(labels))
assert_array_equal(enc.get_classes(), sorted(labels))
assert_array_equal(enc.transform([labels[1], labels[2], "173"]), [2, 1, 3])
# Test that fit_transform has the same behavior
enc = RobustLabelEncoder(labels=labels)
with pytest.warns(UserWarning):
y_transformed = enc.fit_transform([labels[1], labels[2], "173"])
assert_array_equal(list(enc.classes_), sorted(labels))
assert_array_equal(y_transformed, [2, 1, 3])
# Test fill_label_value is not sorted when include_unseen_class is True
enc = RobustLabelEncoder(labels=labels,
fill_label_value="-99",
include_unseen_class=True)
with pytest.warns(UserWarning):
enc.fit([labels[2], labels[0]])
assert_array_equal(enc.get_classes(), sorted(labels) + ["-99"])
def test_robust_label_encoder_error_unknown():
with pytest.raises(ValueError):
enc = RobustLabelEncoder(fill_unseen_labels=False)
enc.fit(X[:, 0])
assert_array_equal(enc.get_classes(), ["apple", "banana", "hot dog"])
enc.transform(["eggplant"])
def test_robust_label_encoder_fill_label_value():
y = np.array([1, 1, 0, 1, 1])
enc = RobustLabelEncoder(labels=[1], fill_label_value=0, include_unseen_class=True)
enc.fit(y)
np.testing.assert_array_equal(enc.get_classes(), [1, 0])
y_transform = enc.transform(y)
np.testing.assert_array_equal(y_transform, [0, 0, 1, 0, 0])
np.testing.assert_array_equal(enc.inverse_transform(y_transform), y)
# Test that fit_transform has the same behavior
enc = RobustLabelEncoder(labels=[1], fill_label_value=0)
y_transform = enc.fit_transform(y)
np.testing.assert_array_equal(enc.get_classes(), [1])
np.testing.assert_array_equal(y_transform, [0, 0, 1, 0, 0])
np.testing.assert_array_equal(enc.inverse_transform(y_transform), y)
def test_robust_label_encoder_sorted_labels(labels):
enc = RobustLabelEncoder(labels=labels)
enc.fit([labels[1], labels[0]])
assert_array_equal(list(enc.classes_), labels)
assert_array_equal(enc.transform([labels[2], labels[1], "173"]), [2, 1, 3])
def test_robust_label_encoder_error_unknown():
with pytest.raises(ValueError):
enc = RobustLabelEncoder(fill_unseen_labels=False)
enc.fit(X[:, 0])
enc.transform(["eggplant"])
def build_label_transform():
"""Returns the model definition representing feature processing."""
return RobustLabelEncoder(labels=['1', '2', '3', '4', '5'])
def build_label_transform():
"""Returns the model definition representing feature processing."""
return RobustLabelEncoder(labels=['0'],
fill_label_value='1',
include_unseen_class=True)
from sagemaker_sklearn_extension.impute import RobustMissingIndicator
from sagemaker_sklearn_extension.preprocessing import LogExtremeValuesTransformer
from sagemaker_sklearn_extension.preprocessing import NALabelEncoder
from sagemaker_sklearn_extension.preprocessing import QuadraticFeatures
from sagemaker_sklearn_extension.preprocessing import QuantileExtremeValuesTransformer
from sagemaker_sklearn_extension.preprocessing import RemoveConstantColumnsTransformer
from sagemaker_sklearn_extension.preprocessing import RobustLabelEncoder
from sagemaker_sklearn_extension.preprocessing import RobustStandardScaler
from sagemaker_sklearn_extension.preprocessing import ThresholdOneHotEncoder
@pytest.mark.parametrize(
"Estimator",
[
DateTimeVectorizer(),
LogExtremeValuesTransformer(),
MultiColumnTfidfVectorizer(),
NALabelEncoder(),
QuadraticFeatures(),
QuantileExtremeValuesTransformer(),
RobustImputer(),
RemoveConstantColumnsTransformer(),
RobustLabelEncoder(),
RobustMissingIndicator(),
RobustStandardScaler(),
ThresholdOneHotEncoder(),
],
)
def test_all_estimators(Estimator):
return check_estimator(Estimator)
