def test_incorrect_test_train_split(dummy_data):
dataloader = BaseDataModule(data=dummy_data,
y_col='y',
train_test_split=42)
with pytest.raises(ValueError):
dataloader.prepare_data()
def test_stratified_shuffle_regression_warning(dummy_data):
with pytest.warns(SunpyUserWarning):
dataloader = BaseDataModule(data=dummy_data,
y_col='y',
is_regression=True,
stratified_shuffle=True,
weighted_sampling=False)
dataloader.prepare_data()
assert dataloader.stratified_shuffle is False
def test_disjoint_split(dummy_data):
dataloader = BaseDataModule(data=dummy_data, y_col='y')
dataloader.prepare_data()
dataloader.setup()
assert set(dataloader.train.index).isdisjoint(dataloader.val.index)
assert set(dataloader.train.index).isdisjoint(dataloader.test.index)
assert set(dataloader.test.index).isdisjoint(dataloader.val.index)
def test_split_distribution(dummy_data):
dataloader = BaseDataModule(data=dummy_data, y_col='y')
dataloader.prepare_data()
dataloader.setup()
data_dist = distribution(dataloader.data['y'])
assert not np.any(
np.abs(distribution(dataloader.train['y']) - data_dist) > 0.1)
assert not np.any(
np.abs(distribution(dataloader.test['y']) - data_dist) > 0.1)
assert not np.any(
np.abs(distribution(dataloader.val['y']) - data_dist) > 0.1)
def test_oversampling(dummy_data):
train_conf = {'is_tabular': True, 'transform': ToTensor()}
dataloader = BaseDataModule(data=dummy_data,
y_col='y',
batch_size=1,
train_conf=train_conf)
dataloader.prepare_data()
dataloader.setup()
sampled_labels = []
for index, sample in enumerate(dataloader.train_dataloader()):
sampled_labels.append(sample[1].numpy()[0])
data_distribution = distribution(sampled_labels)
assert np.abs(data_distribution[0] - data_distribution[1]) < 0.2
def test_conf_passed(dummy_data):
training_conf = {}
testing_conf = {}
validation_conf = {}
dataloader = BaseDataModule(data=dummy_data,
y_col='y',
train_conf=training_conf,
test_conf=testing_conf,
val_conf=validation_conf)
dataloader.prepare_data()
with pytest.warns(None) as record:
dataloader.setup()
assert not record
def test_no_conf(dummy_data):
training_conf = {}
testing_conf = {}
validation_conf = {}
dataloader_no_train_conf = BaseDataModule(data=dummy_data,
y_col='y',
test_conf=testing_conf,
val_conf=validation_conf)
dataloader_no_train_conf.prepare_data()
with pytest.warns(SunpyUserWarning):
dataloader_no_train_conf.setup()
dataloader_no_val_conf = BaseDataModule(data=dummy_data,
y_col='y',
test_conf=testing_conf,
train_conf=training_conf)
dataloader_no_val_conf.prepare_data()
with pytest.warns(SunpyUserWarning):
dataloader_no_val_conf.setup()
dataloader_no_test_conf = BaseDataModule(data=dummy_data,
y_col='y',
train_conf=training_conf,
val_conf=validation_conf)
dataloader_no_test_conf.prepare_data()
with pytest.warns(SunpyUserWarning):
dataloader_no_test_conf.setup()
def test_incorrect_data():
dataloader = BaseDataModule(data=42, y_col='y')
with pytest.raises(TypeError):
dataloader.prepare_data()
def test_default_loader(dummy_data):
BaseDataModule(data=dummy_data, y_col='y')
def test_X_y_disjoint(dummy_data):
with pytest.raises(ValueError):
dataloader = BaseDataModule(data=dummy_data,
y_col='y',
X_col=['X', 'y'])
def test_no_X_col(dummy_data):
with pytest.warns(SunpyUserWarning):
dataloader = BaseDataModule(data=dummy_data, y_col='y')
dataloader.prepare_data()