def calc_stats(test_label, results_file, human_suggestions_file, stats_file):
"""Calc stats and write to results_file.
Arguments:
- `query_list`:
- `suggestion_dict`:
"""
query_list, suggestion_dict = get_output_from_file(test_label, results_file)
human_suggestion_dict = get_human_suggestions(test_label, human_suggestions_file)
dummy_spell_checker = spell_checker.SpellChecker()
args = [query_list, suggestion_dict, human_suggestion_dict]
EP = utils.get_EP(*args)
ER = utils.get_ER(*args)
EF1 = utils.get_HM(EP, ER)
stats = [EP, ER, EF1]
# print 'stats', stats
f = open(stats_file, 'a')
stats_str = 'Timestamp: {0}\tLabel: {1}\tEP: {3}\tER: {4}\t \
EF1: {5}\tNum queries: {2}\n'.format(
str(datetime.now()), test_label, len(query_list), *stats)
print 'stats_str', stats_str
f.write(stats_str)
f.close()
def get_EF1_measure(self, human_suggestion_dict):
"""Return EF1 value for the performance as judged by human_suggestion_dict.
Arguments:
- `human_suggestion_dict`: dict of query -> list of
human-annotated suggestions
"""
args = [self.query_list, self.suggestion_dict, human_suggestion_dict]
return utils.get_HM (utils.get_EP(*args),
utils.get_ER(*args))
def test_get_HM_corner_cases(self):
self.assertAlmostEqual(utils.get_HM(0.0, 3.0), 0.0, 1)
self.assertAlmostEqual(utils.get_HM(0.3, 0.0), 0.0, 1)
self.assertAlmostEqual(utils.get_HM(0.0, 0.0), 0.0, 1)
def test_get_HM(self):
self.assertAlmostEqual(utils.get_HM(3, 3), 3, 1)
self.assertAlmostEqual(utils.get_HM(3, 0), 0, 1)
self.assertAlmostEqual(utils.get_HM(0.3, 0.7), 0.42, 1)
self.assertAlmostEqual(utils.get_HM(0.5, 0.5), 0.5, 1)
self.assertAlmostEqual(utils.get_HM(0.1, 0.9), 0.18, 1)
