def setUp(self):
self.X, self.y = rand_df(shape=(100, 10))
self.test_X = rand_df(shape=(100, 10), labeled=False)
self.steps = [
EmptyStep(),
PearsonCorrStep(4),
PCAStep(kwargs={'n_components': 2})
]
def setUp(self):
self.X = rand_df(shape=(100, 10), labeled=False)
self.test_X = rand_df(shape=(100, 10), labeled=False)
self.steps = [
EmptyStep(),
StandardScalerStep(append_input=True),
PCAStep(kwargs={'n_components': 2})
]
class ListTests(unittest.TestCase, StepTest):
step = ListSelectionStep(features=['11', '22'])
X, y = rand_df()
test_X = rand_df(labeled=False)
# Tests that a key error is raised when the features do not exist
def test_key_error(self):
s = ListSelectionStep(features=['11', 'not_a_feature'])
tr_X, tr_y = rand_df(shape=(10, 15))
with self.assertRaises(KeyError):
s.fit(tr_X, y=tr_y)
def test_pipeline_step(self):
tr_data_X = rand_df(shape=(100, 20), labeled=False)
te_data = rand_df(shape=(100, 20), labeled=False)
scale_step = StandardScalerStep()
pca_step = PCAStep(append_input=False, kwargs={'n_components': 5})
poly_step = PolyStep(kwargs={'degree': 3, 'include_bias': False})
pipeline = Pipeline([
scale_step,
Pipeline([pca_step, poly_step], append_input=True),
EmptyStep()
])
r = pipeline.fit_transform(tr_data_X)
self.assertEqual(type(r), pd.DataFrame)
r = pipeline.transform(te_data)
self.assertEqual(type(r), pd.DataFrame)
def test_pipeline_step(self):
tr_data_X, tr_data_y = rand_df_classification(shape=(100, 20),
classes=3)
te_data = rand_df(shape=(100, 20), labeled=False)
scale_step = StandardScalerStep()
chi_step = ChiSqSelectionStep(select_kwargs={'k': 20})
corr_step = PearsonCorrStep(num_features=0.1)
pca_step = PCAStep(append_input=False, kwargs={'n_components': 5})
poly_step = PolyStep(kwargs={'degree': 3, 'include_bias': False})
pipeline = Pipeline([
scale_step,
Pipeline([pca_step, poly_step], append_input=True), chi_step
])
r, _ = pipeline.fit_transform(tr_data_X, y=tr_data_y)
self.assertEqual(type(r), pd.DataFrame)
r = pipeline.transform(te_data)
self.assertEqual(type(r), pd.DataFrame)
class ADASYNTests1(unittest.TestCase, StepTest):
step = ADASYNStep(kwargs={'ratio': {0.0: 100, 1.0: 100}})
X, y = rand_df_classification(val_range=(0, 100))
test_X = rand_df(val_range=(0, 100))
class SinTests3(unittest.TestCase, StepTest):
step = SinStep(append_input=True, columns=['1', '2', '3'])
X, y = rand_df()
test_X = rand_df(labeled=False)
class SinTests1(unittest.TestCase, StepTest):
step = SinStep()
X, y = rand_df()
test_X = rand_df(labeled=False)
class PolyTests1(unittest.TestCase, StepTest):
step = PolyStep()
X, y = rand_df()
test_X = rand_df(labeled=False)
class LogTests3(unittest.TestCase, StepTest):
step = LogStep(append_input=True, columns=['1', '10'], log_func=np.log10)
X, y = rand_df(val_range=(0, 100))
test_X = rand_df(val_range=(0, 100), labeled=False)
class LogTests1(unittest.TestCase, StepTest):
step = LogStep()
X, y = rand_df(val_range=(0, 100))
test_X = rand_df(val_range=(0, 100), labeled=False)
class TreeTests2(unittest.TestCase, StepTest):
step = TreeSelectionStep(tree_model=ExtraTreesClassifier)
X, y = rand_df_classification()
test_X = rand_df(labeled=False)
class TreeTests1(unittest.TestCase, StepTest):
step = TreeSelectionStep()
X, y = rand_df()
test_X = rand_df(labeled=False)
class PearsonCorrTests2(unittest.TestCase, StepTest):
step = PearsonCorrStep(num_features=50)
X, y = rand_df()
test_X = rand_df(labeled=False)
class ListTests2(unittest.TestCase, StepTest):
step = ListSelectionStep(features=['11', '22'])
X = rand_df(labeled=False)
y = None
test_X = rand_df(labeled=False)
class LDATransform1(unittest.TestCase, StepTest):
step = LDATransformStep()
X, y = rand_df_classification()
test_X = rand_df(labeled=False)
class LDATransform2(unittest.TestCase, StepTest):
step = LDATransformStep(append_input=True)
X, y = rand_df_classification()
test_X = rand_df(labeled=False)
class ABODTests(unittest.TestCase, StepTest):
step = ABODStep(num_remove=1)
X, y = rand_df_classification(outlier=True)
test_X = rand_df(labeled=False, outlier=True)
class LogTests2(unittest.TestCase, StepTest):
step = LogStep()
X = rand_df(val_range=(0, 100), labeled=False)
y = None
test_X = rand_df(val_range=(0, 100), labeled=False)
class IsoForestDefault(unittest.TestCase, StepTest):
step = IsoForestStep()
X, y = rand_df(outlier=True)
test_X = rand_df(labeled=False, outlier=True)
class PCATests2(unittest.TestCase, StepTest):
step = PCAStep(append_input=True)
X = rand_df(labeled=False)
y = None
test_X = rand_df(labeled=False)
class IsoForestIncludeY(unittest.TestCase, StepTest):
step = IsoForestStep(include_y=False)
X, y = rand_df(outlier=True)
test_X = rand_df(labeled=False, outlier=True)
class PolyTests3(unittest.TestCase, StepTest):
step = PolyStep(append_input=True)
X = rand_df(shape=(100, 10), labeled=False)
y = None
test_X = rand_df(shape=(50, 10), labeled=False)
class LFODefault(unittest.TestCase, StepTest):
step = LOFStep()
X, y = rand_df_classification(outlier=True)
test_X = rand_df(labeled=False, outlier=True)
class SinTests2(unittest.TestCase, StepTest):
step = SinStep()
X = rand_df(labeled=False)
y = None
test_X = rand_df(labeled=False)
class LFOIncludeY(unittest.TestCase, StepTest):
step = LOFStep(include_y=False)
X, y = rand_df(outlier=True)
test_X = rand_df(labeled=False, outlier=True)
class StandardScalerTests1(unittest.TestCase, StepTest):
step = StandardScalerStep()
X, y = rand_df()
test_X = rand_df(labeled=False)
class StandardScalerTests2(unittest.TestCase, StepTest):
step = StandardScalerStep()
X = rand_df(labeled=False)
y = None
test_X = rand_df(labeled=False)
class SMOTETests1(unittest.TestCase, StepTest):
step = SMOTEStep()
X, y = rand_df_classification(val_range=(0, 100))
test_X = rand_df(val_range=(0, 100))
class StandardScalerTests3(unittest.TestCase, StepTest):
step = StandardScalerStep(append_input=True)
X, y = rand_df()
test_X = rand_df(labeled=False)
