| 47 | |
| 48 | |
| 49 | class TestUtils(unittest.TestCase): |
| 50 | |
| 51 | def test_check_class_labels(self): |
| 52 | for n_labels in range(1, 10): |
| 53 | for n_learners in range(1, 10): |
| 54 | # 1. test fitted estimators |
| 55 | labels = np.random.randint(10, size=n_labels) |
| 56 | different_labels = np.random.randint( |
| 57 | 10, 20, size=np.random.randint(1, 10)) |
| 58 | learner_list_1 = [mock.MockEstimator( |
| 59 | classes_=labels) for _ in range(n_learners)] |
| 60 | learner_list_2 = [mock.MockEstimator( |
| 61 | classes_=different_labels) for _ in range(np.random.randint(1, 5))] |
| 62 | shuffled_learners = random.sample( |
| 63 | learner_list_1 + learner_list_2, len(learner_list_1 + learner_list_2)) |
| 64 | self.assertTrue( |
| 65 | modAL.utils.validation.check_class_labels(*learner_list_1)) |
| 66 | self.assertFalse( |
| 67 | modAL.utils.validation.check_class_labels(*shuffled_learners)) |
| 68 | |
| 69 | # 2. test unfitted estimators |
| 70 | unfitted_learner_list = [mock.MockEstimator( |
| 71 | classes_=labels) for _ in range(n_learners)] |
| 72 | idx = np.random.randint(0, n_learners) |
| 73 | unfitted_learner_list.insert( |
| 74 | idx, mock.MockEstimator(fitted=False)) |
| 75 | self.assertRaises( |
| 76 | NotFittedError, modAL.utils.validation.check_class_labels, *unfitted_learner_list) |
| 77 | |
| 78 | def test_check_class_proba(self): |
| 79 | for n_labels in range(2, 20): |
| 80 | # when all classes are known: |
| 81 | proba = np.random.rand(100, n_labels) |
| 82 | class_labels = list(range(n_labels)) |
| 83 | np.testing.assert_almost_equal( |
| 84 | modAL.utils.check_class_proba( |
| 85 | proba, known_labels=class_labels, all_labels=class_labels), |
| 86 | proba |
| 87 | ) |
| 88 | for unknown_idx in range(n_labels): |
| 89 | all_labels = list(range(n_labels)) |
| 90 | known_labels = deepcopy(all_labels) |
| 91 | known_labels.remove(unknown_idx) |
| 92 | aug_proba = np.insert( |
| 93 | proba[:, known_labels], unknown_idx, np.zeros(len(proba)), axis=1) |
| 94 | np.testing.assert_almost_equal( |
| 95 | modAL.utils.check_class_proba( |
| 96 | proba[:, known_labels], known_labels=known_labels, all_labels=all_labels), |
| 97 | aug_proba |
| 98 | ) |
| 99 | |
| 100 | def test_linear_combination(self): |
| 101 | |
| 102 | def dummy_function(X_in): |
| 103 | return np.ones(shape=(len(X_in), 1)) |
| 104 | |
| 105 | for n_samples in range(2, 10): |
| 106 | for n_features in range(1, 10): |
nothing calls this directly
no outgoing calls
no test coverage detected
searching dependent graphs…