def test_input_shape_tabular_valid(self):
test_num_samples = [1, 2, 1024]
test_num_features = [1, 2, 1024]
for num_samples in test_num_samples:
for num_features in test_num_features:
x = np.random.random_sample(size=(num_samples, num_features))
n, input_dim = Validation.get_input_dimension(x)
self.assertEqual(n, num_samples)
self.assertEqual(input_dim, (num_features,))
def test_input_shape_time_series_variable_valid(self):
test_num_samples = [2, 3, 1024]
test_num_lens = [1, 2, 256]
test_num_features = [1, 2, 1024]
for num_samples in test_num_samples:
for num_features in test_num_features:
x = [np.random.random_sample(size=(test_num_lens[i % len(test_num_lens)], num_features))
for i in range(num_samples)]
n, input_dim = Validation.get_input_dimension(x)
self.assertEqual(n, num_samples)
self.assertEqual(input_dim, (None, num_features))
def test_input_shape_time_series_fixed_valid(self):
test_num_samples = [1, 2, 1024]
test_num_lens = [1, 2, 256]
test_num_features = [1, 2, 1024]
for num_samples in test_num_samples:
for ts_length in test_num_lens:
for num_features in test_num_features:
x = np.random.random_sample(size=(num_samples, ts_length, num_features))
n, input_dim = Validation.get_input_dimension(x)
self.assertEqual(n, num_samples)
self.assertEqual(input_dim, (ts_length, num_features))
def test_input_shape_invalid_1dim(self):
with self.assertRaises(ValueError):
Validation.get_input_dimension([1])
with self.assertRaises(ValueError):
Validation.get_input_dimension([1, 2, 3])
with self.assertRaises(ValueError):
Validation.get_input_dimension([None])
def test_input_shape_image_fixed_valid(self):
test_num_samples = [1, 2, 1024]
test_num_rows = [1, 2, 256]
test_num_cols = [1, 2, 256]
test_num_channels = [1, 2, 3]
for num_samples in test_num_samples:
for rows in test_num_rows:
for cols in test_num_cols:
for num_channels in test_num_channels:
x = np.random.random_sample(size=(num_samples, rows, cols, num_channels))
n, input_dim = Validation.get_input_dimension(x)
self.assertEqual(n, num_samples)
self.assertEqual(input_dim, (rows, cols, num_channels))
def test_input_shape_volume_fixed_valid(self):
test_num_samples = [1, 2, 128]
test_num_voxels = [1, 2, 64]
test_num_channels = [1, 2, 3]
for num_samples in test_num_samples:
for rows in test_num_voxels:
for cols in test_num_voxels:
for depth in test_num_voxels:
for num_channels in test_num_channels:
x = np.random.random_sample(size=(num_samples, rows, cols, depth, num_channels))
n, input_dim = Validation.get_input_dimension(x)
self.assertEqual(n, num_samples)
self.assertEqual(input_dim, (rows, cols, depth, num_channels))
def test_input_shape_volume_variable_valid(self):
test_num_samples = [2, 3, 128]
test_num_lens = [2, 3, 64]
test_num_features = [1, 2, 3]
for num_samples in test_num_samples:
for num_features in test_num_features:
x = [np.random.random_sample(size=(test_num_lens[i % len(test_num_lens)],
test_num_lens[(i + 1) % len(test_num_lens)],
test_num_lens[(i + 2) % len(test_num_lens)],
num_features))
for i in range(num_samples)]
n, input_dim = Validation.get_input_dimension(x)
self.assertEqual(n, num_samples)
self.assertEqual(input_dim, (None, None, None, num_features))
def _build_model(self, X, y):
Validation.check_dataset(X, y)
if Validation.is_variable_length(X):
raise ValueError("Variable length inputs to CXPlain are currently not supported.")
n, p = Validation.get_input_dimension(X)
output_dim = Validation.get_output_dimension(y)
if self.model is None:
if self.num_models == 1:
build_fun = self._build_single
else:
build_fun = self._build_ensemble
self.model, self.prediction_model = build_fun(input_dim=p, output_dim=output_dim)
def test_input_shape_invalid_none(self):
with self.assertRaises(ValueError):
Validation.get_input_dimension(None)