def i_make_a_prediction(step, model, text, expected):
topic_model = TopicModel(model)
distribution = topic_model.distribution(text)
msg = ("Computed distribution is %s, but expected distribution is %s" %
(str(distribution), str(expected)))
eq_(len(distribution), len(expected), msg)
for d, e in zip(distribution, expected):
assert_almost_equals(d['probability'],
e['probability'],
places=6, msg=msg)
def local_topic_distribution(topic_models,
test_reader,
output,
args,
exclude=None,
headers=None):
"""Get local topic model and issue topic distribution prediction
"""
# Only one topic model at present
local_topic_model = TopicModel(topic_models[0], api=args.retrieve_api_)
if args.prediction_header:
headers.extend([topic['name'] for topic in local_topic_model.topics])
output.writerow(headers)
for input_data in test_reader:
input_data_dict = test_reader.dict(input_data, filtering=False)
try:
topic_distribution_info = local_topic_model.distribution(
input_data_dict)
except Exception:
topic_distribution_info = []
write_topic_distribution(topic_distribution_info, output,
args.prediction_info, input_data, exclude)
def i_create_local_topic_model_from_file(step, export_file):
world.local_topic_model = TopicModel(res_filename(export_file))
def i_create_a_local_topic_model(step):
world.local_topic_model = TopicModel(world.topic_model)