def test_sparse_vector(self):
maxent = MaxEnt()
_ = maxent.compute_observed_counts(self.training_set)
woof_index = maxent.feature_alphabet.get_index('woof')
meow_index = maxent.feature_alphabet.get_index('meow')
sparse_vector = self.training_set[1].data
self.assertEqual(meow_index, sparse_vector[0])
self.assertEqual(woof_index, sparse_vector[1])
def test_label_loglikelihood_score(self): maxent = MaxEnt() _ = maxent.compute_observed_counts(self.training_set) maxent.parameters = numpy.array([0.1, 0.5, 1, 0, 100, -0.1, -0.5, -1, 0, -100]) score_vector = maxent.compute_label_unnormalized_loglikelihood_vector(self.training_set[0].data) self.assertEqual(score_vector[0], 1.6) self.assertEqual(score_vector[1], -1.6) posterior_distribution = maxent.compute_posterior_distribution(self.training_set[0]) self.assertAlmostEqual(posterior_distribution[0], 0.96, 2)
def test_observed_counts(self):
maxent = MaxEnt()
observed_counts = maxent.compute_observed_counts(self.training_set)
cat_index = maxent.label_alphabet.get_index('cat')
purr_index = maxent.feature_alphabet.get_index('purr')
meow_index = maxent.feature_alphabet.get_index('meow')
bark_index = maxent.feature_alphabet.get_index('bark')
woof_index = maxent.feature_alphabet.get_index('woof')
cat_related_indices = maxent.get_parameter_indices([purr_index, meow_index, bark_index, woof_index], cat_index)
cat_count, cat_purr_count, cat_meow_count, cat_bark_count, cat_woof_count = observed_counts[cat_related_indices]
self.assertEqual(cat_count, 3)
self.assertEqual(cat_purr_count, 2)
self.assertEqual(cat_meow_count, 2)
self.assertEqual(cat_bark_count, 0)
self.assertEqual(cat_woof_count, 1)
