def testPartiallySpecifiedTestSet(self):
"""Check that partially specified test set raises an error."""
num_features = 5
num_test_points = 5
dataset = _test_dataset(num_features, num_test_points)
del dataset['test_features']
with self.assertRaisesRegex(ValueError, 'both specified'):
sparse_logistic_regression.SparseLogisticRegression(**dataset)
def testBasic(self, num_test_points):
"""Checks that you get finite values given unconstrained samples.
We check `unnormalized_log_prob` as well as the values of the sample
transformations.
Args:
num_test_points: Number of test points.
"""
num_features = 5
model = sparse_logistic_regression.SparseLogisticRegression(
**_test_dataset(num_features, num_test_points))
self.validate_log_prob_and_transforms(
model,
sample_transformation_shapes=dict(
identity={
'global_scale': [],
'local_scales': [num_features + 1],
'unscaled_weights': [num_features + 1],
},
test_nll=[],
per_example_test_nll=[num_test_points],
))