def test_estimate_competence_kuncheva_ex(example_kuncheva):
query = np.array([1, 1])
a_priori_test = APriori(k=example_kuncheva['k'])
test_example = example_kuncheva
a_priori_test.DSEL_processed_ = test_example['dsel_processed']
a_priori_test.dsel_scores_ = test_example['dsel_scores']
a_priori_test.DSEL_target_ = test_example['y_independent']
a_priori_test.n_classes_ = test_example['n_classes']
neighbors = test_example['neighbors'].reshape(1, -1)
distances = test_example['distances'].reshape(1, -1)
competences = a_priori_test.estimate_competence(query.reshape(1, -1),
neighbors, distances)
assert np.isclose(competences, 0.70, atol=0.01)
def test_estimate_competence_all_ones(index, expected):
query = np.array([1, 1])
a_priori_test = APriori(create_pool_classifiers())
a_priori_test.DSEL_processed_ = dsel_processed_ex1
a_priori_test.dsel_scores_ = dsel_scores_all_ones
a_priori_test.DSEL_target_ = y_dsel_ex1
a_priori_test.n_classes_ = 2
neighbors = neighbors_ex1[index, :].reshape(1, -1)
distances = distances_all_ones[index, :].reshape(1, -1)
competences = a_priori_test.estimate_competence(query.reshape(1, -1),
neighbors, distances)
assert np.isclose(competences, expected).all()
def test_estimate_competence_kuncheva_ex():
query = np.array([1, 1])
a_priori_test = APriori([create_base_classifier(return_value=1)],
k=k_ex_kuncheva)
a_priori_test.DSEL_processed_ = dsel_processed_kuncheva
a_priori_test.dsel_scores_ = dsel_scores_ex_kuncheva
a_priori_test.DSEL_target_ = y_dsel_ex_kuncheva_independent
a_priori_test.n_classes_ = n_classes_ex_kuncheva
neighbors = neighbors_ex_kuncheva.reshape(1, -1)
distances = distances_ex_kuncheva.reshape(1, -1)
competences = a_priori_test.estimate_competence(query.reshape(1, -1),
neighbors, distances)
assert np.isclose(competences, 0.70, atol=0.01)
def test_estimate_competence_all_ones(index, expected, example_all_ones):
X, y, neighbors, distances, dsel_processed, dsel_scores = example_all_ones
query = np.array([1, 1])
a_priori_test = APriori()
a_priori_test.DSEL_processed_ = dsel_processed
a_priori_test.dsel_scores_ = dsel_scores
a_priori_test.DSEL_target_ = y
a_priori_test.n_classes_ = 2
neighbors = neighbors[index, :].reshape(1, -1)
distances = distances[index, :].reshape(1, -1)
competences = a_priori_test.estimate_competence(query.reshape(1, -1),
neighbors, distances)
assert np.isclose(competences, expected).all()
def test_estimate_competence_batch(example_estimate_competence):
_, y, nn, _, dsel_processed, dsel_scores = example_estimate_competence
expected = np.array([[0.333333, 0.50000, 0.40000],
[0.666666, 0.50000, 0.60000],
[0.000000, 0.50000, 0.20000]])
# Using 3 neighbors to facilitate the calculations
a_priori_test = APriori(k=3)
a_priori_test.DSEL_processed_ = dsel_processed
a_priori_test.dsel_scores_ = dsel_scores
a_priori_test.DSEL_target_ = y
a_priori_test.n_classes_ = 2
nn = nn[:, 0:3]
distances = np.ones((3, 3))
competences = a_priori_test.estimate_competence(nn,
distances)
assert np.allclose(competences, expected, atol=0.01)
def test_estimate_competence_batch():
query = np.ones((3, 2))
expected = np.array([[0.333333, 0.50000, 0.40000],
[0.666666, 0.50000, 0.60000],
[0.000000, 0.50000, 0.20000]])
# Using 3 neighbors to facilitate the calculations
a_priori_test = APriori(create_pool_classifiers(), 3)
a_priori_test.DSEL_processed_ = dsel_processed_ex1
a_priori_test.dsel_scores_ = dsel_scores_ex1
a_priori_test.DSEL_target_ = y_dsel_ex1
a_priori_test.n_classes_ = 2
neighbors = neighbors_ex1[:, 0:3]
distances = distances_all_ones[:, 0:3]
competences = a_priori_test.estimate_competence(query, neighbors,
distances)
assert np.allclose(competences, expected, atol=0.01)