def test_creation_from_file(self):
"""Test that a Corpus instance is properly created from a file."""
file_path = os.path.join(self.tempdir.name, "file")
with open(file_path, "w") as wf:
for document in self.list_of_documents:
line = " ".join(document) + "\n"
wf.write(line)
corpus_from_file = Corpus.from_document_file(file_path)
corpus_from_documents = Corpus.from_iterable_of_word_lists(self.list_of_documents)
self.assertEqual(corpus_from_documents, corpus_from_file, "Loaded corpus is not correct.")
def setUp(self):
"""Code to run before every test."""
self.list_of_documents = [
["the", "quick", "brown", "fox"],
["jumped", "over"],
["the", "lazy", "lazy", "dog"]
]
self.corpus = Corpus.from_iterable_of_word_lists(self.list_of_documents)
def setUp(self):
"""Code to run before each test."""
corpus = Corpus.from_document_file("tests/data/sample_data")
self.model = GibbsSamplingDMM(corpus, number_of_topics=20)
self.model.randomly_initialise_topic_assignment(seed=1)
self.model.inference(5)
self.tempdir = tempfile.TemporaryDirectory()
def setUp(self):
"""Code to run before each test."""
corpus = Corpus.from_document_file("tests/data/sample_data")
self.model = GibbsSamplingDMM(corpus, number_of_topics=20)
self.model.randomly_initialise_topic_assignment(seed=1)
self.model.inference(100)
self.generated_documents, self.chosen_topics = self.model.generate_synthetic_documents(
10, seed=5)
def test_speed_of_inference(self):
"""Test that the inference is fast enough."""
corpus = Corpus.from_document_file("tests/data/sample_data")
model = GibbsSamplingDMM(corpus, number_of_topics=20)
model.randomly_initialise_topic_assignment(seed=1)
number_of_iterations = 200
t0 = time.time()
model.inference(number_of_iterations)
t1 = time.time()
average_seconds_per_iteration = (t1 - t0) / number_of_iterations
expected_seconds_per_iteration = 0.02
delta = 0.05
if average_seconds_per_iteration < expected_seconds_per_iteration - delta:
self.fail(
"Code was faster: actually took {:.5f} per iteration.".format(
average_seconds_per_iteration))
else:
self.assertAlmostEqual(average_seconds_per_iteration,
expected_seconds_per_iteration,
delta=0.05)
def test_bad_equality(self):
"""Test that a Corpus instance is not equal to an integer."""
corpus = Corpus.from_iterable_of_word_lists(self.list_of_documents)
with self.assertRaises(TypeError, msg="Comparison to incorrect type should raise an error."):
bool(corpus == 5)
def setUp(self):
"""Code to run before each test."""
corpus = Corpus.from_document_file("tests/data/sample_data")
self.model = GibbsSamplingDMM(corpus, number_of_topics=20)
self.model.randomly_initialise_topic_assignment(seed=1)