def test_empty_pool(create_X_y):
pool_classifiers = []
X, y = create_X_y
with pytest.raises(ValueError):
ds = BaseDS(pool_classifiers)
ds.fit(X, y)
def test_preprocess_dsel_scores():
ds_test = BaseDS(create_pool_classifiers())
ds_test.fit(X_dsel_ex1, y_dsel_ex1)
dsel_scores = ds_test._preprocess_dsel_scores()
expected = np.array([[0.5, 0.5], [1.0, 0.0], [0.33, 0.67]])
expected = np.tile(expected, (15, 1, 1))
assert np.array_equal(dsel_scores, expected)
def test_frienemy_safe_region():
ds_test = BaseDS(create_pool_classifiers(), safe_k=3)
ds_test.fit(X_dsel_ex1, y_dsel_ex1)
ds_test.DSEL_processed_ = dsel_processed_ex1
result = ds_test._frienemy_pruning(np.array([0, 1, 2, 6, 7, 8, 14]))
assert np.array_equal(result, np.array([[1, 1, 1]]))
def test_frienemy_not_all_classifiers_crosses():
ds_test = BaseDS(create_pool_classifiers(), safe_k=3)
ds_test.fit(X_dsel_ex1, y_dsel_ex1)
ds_test.DSEL_processed_ = dsel_processed_ex1
result = ds_test._frienemy_pruning(neighbors_ex1[0, :])
assert np.array_equal(result, np.array([[1, 1, 0]]))
def test_frienemy_no_classifier_crosses():
X = X_dsel_ex1
y = y_dsel_ex1
ds_test = BaseDS(create_pool_classifiers())
ds_test.fit(X, y)
mask = ds_test._frienemy_pruning(neighbors_ex1[0, :])
assert mask.shape == (1, 3) and np.allclose(mask, 1)
def test_frienemy_all_classifiers_crosses(index):
ds_test = BaseDS(create_pool_classifiers())
ds_test.fit(X_dsel_ex1, y_dsel_ex1)
ds_test.DSEL_processed_ = dsel_processed_all_ones
result = ds_test._frienemy_pruning(neighbors_ex1[index, :])
assert result.all() == 1.0
def test_empty_pool():
pool_classifiers = []
X = np.random.rand(10, 2)
y = np.ones(10)
with pytest.raises(ValueError):
ds = BaseDS(pool_classifiers)
ds.fit(X, y)
def test_check_k_type(k):
X = np.random.rand(10, 2)
y = np.ones(10)
with pytest.raises(TypeError):
ds_test = BaseDS(k=k)
ds_test.fit(X, y)
def test_different_input_shape(create_X_y):
X, y = create_X_y
query = np.array([[1.0, 1.0, 2.0]])
ds_test = BaseDS()
ds_test.fit(X, y)
with pytest.raises(ValueError):
ds_test.predict(query)
def test_check_safe_k_value(safe_k):
pool_classifiers = create_pool_classifiers()
X = np.random.rand(10, 2)
y = np.ones(10)
with pytest.raises(ValueError):
ds_test = BaseDS(pool_classifiers, safe_k=safe_k)
ds_test.fit(X, y)
def test_check_k_type(k):
pool_classifiers = create_pool_classifiers()
X = np.random.rand(10, 2)
y = np.ones(10)
with pytest.raises(TypeError):
ds_test = BaseDS(pool_classifiers, k=k)
ds_test.fit(X, y)
def test_frienemy_no_classifier_crosses(example_estimate_competence,
create_pool_classifiers):
X, y, neighbors = example_estimate_competence[0:3]
ds_test = BaseDS(create_pool_classifiers)
ds_test.fit(X, y)
mask = ds_test._frienemy_pruning(neighbors[0, :])
assert mask.shape == (1, 3) and np.allclose(mask, 1)
def test_DFP_is_used(example_estimate_competence, create_pool_classifiers):
X, y, neighbors, _, dsel_processed, _ = example_estimate_competence
ds_test = BaseDS(create_pool_classifiers, DFP=True, safe_k=3)
ds_test.fit(X, y)
ds_test.DSEL_processed_ = dsel_processed
DFP_mask = ds_test._frienemy_pruning(neighbors[0, :])
assert np.array_equal(DFP_mask, np.atleast_2d([1, 1, 0]))
def test_preprocess_dsel_scores(create_X_y, create_pool_classifiers):
X, y = create_X_y
ds_test = BaseDS(create_pool_classifiers)
ds_test.fit(X, y)
dsel_scores = ds_test._predict_proba_base(X)
expected = np.array([[0.5, 0.5], [1.0, 0.0], [0.33, 0.67]])
expected = np.tile(expected, (15, 1, 1))
assert np.array_equal(dsel_scores, expected)
def test_DFP_is_used():
ds_test = BaseDS(create_pool_classifiers(), DFP=True, safe_k=3)
ds_test.fit(X_dsel_ex1, y_dsel_ex1)
ds_test.DSEL_processed_ = dsel_processed_ex1
ds_test.DSEL_target_ = y_dsel_ex1
ds_test.DSEL_data_ = X_dsel_ex1
DFP_mask = ds_test._frienemy_pruning(neighbors_ex1[0, :])
assert np.array_equal(DFP_mask, np.atleast_2d([1, 1, 0]))
def test_import_faiss_mode():
try:
import sys
sys.modules.pop('deslib.util.faiss_knn_wrapper')
except Exception:
pass
with unittest.mock.patch.dict('sys.modules', {'faiss': None}):
with pytest.raises(ImportError):
BaseDS(knn_classifier="faiss")
def test_frienemy_safe_region(example_estimate_competence,
create_pool_classifiers):
X, y, _, _, dsel_processed, _ = example_estimate_competence
ds_test = BaseDS(create_pool_classifiers, safe_k=3)
ds_test.fit(X, y)
ds_test.DSEL_processed_ = dsel_processed
result = ds_test._frienemy_pruning(np.array([0, 1, 2, 6, 7, 8, 14]))
assert np.array_equal(result, np.array([[1, 1, 1]]))
def test_predict_value(query):
pool_classifiers = create_classifiers_disagree()
ds = BaseDS(pool_classifiers)
X = np.random.rand(10, 2)
y = np.ones(10)
y[:5] = 0
ds.fit(X, y)
with pytest.raises(ValueError):
ds.predict(query)
def test_frienemy_all_classifiers_crosses(index, example_all_ones,
create_pool_classifiers):
X, y, neighbors, _, dsel_processed, _ = example_all_ones
ds_test = BaseDS(create_pool_classifiers)
ds_test.fit(X, y)
ds_test.DSEL_processed_ = dsel_processed
result = ds_test._frienemy_pruning(neighbors[index, :])
assert result.all() == 1.0
def test_predict_proba_all_agree(example_estimate_competence,
create_pool_all_agree):
X, y, _, _, _, dsel_scores = example_estimate_competence
query = np.atleast_2d([1, 1])
ds_test = BaseDS(create_pool_all_agree)
ds_test.fit(X, y)
ds_test.DSEL_scores = dsel_scores
proba = ds_test.predict_proba(query)
assert np.allclose(proba, np.atleast_2d([0.61, 0.39]))
def test_label_encoder_only_dsel_allagree():
X_dsel_ex1 = np.array([[-1, 1], [-0.75, 0.5], [-1.5, 1.5]])
y_dsel_ex1 = np.array(['cat', 'dog', 'plane'])
query = np.atleast_2d([[1, 0], [-1, -1]])
ds_test = BaseDS(create_pool_classifiers_dog(), k=2)
ds_test.fit(X_dsel_ex1, y_dsel_ex1)
predictions = ds_test.predict(query)
assert np.array_equal(predictions, ['dog', 'dog'])
def test_frienemy_not_all_classifiers_crosses_batch():
expected = np.array([[1, 1, 0], [0, 1, 0], [1, 1, 1]])
ds_test = BaseDS(create_pool_classifiers(), safe_k=3)
ds_test.fit(X_dsel_ex1, y_dsel_ex1)
ds_test.DSEL_processed_ = dsel_processed_ex1
# passing three samples to compute the DFP at the same time
result = ds_test._frienemy_pruning(neighbors_ex1)
assert np.array_equal(result, expected)
def test_predict_proba_all_agree():
query = np.atleast_2d([1, 1])
ds_test = BaseDS(create_pool_classifiers())
ds_test.fit(X_dsel_ex1, y_dsel_ex1)
ds_test.DSEL_scores = dsel_scores_ex1
backup_all_agree = BaseDS._all_classifier_agree
BaseDS._all_classifier_agree = MagicMock(return_value=np.array([True]))
proba = ds_test.predict_proba(query)
BaseDS._all_classifier_agree = backup_all_agree
assert np.allclose(proba, np.atleast_2d([0.61, 0.39]))
def test_predict_proba_instance_called(index):
query = np.atleast_2d([1, 1])
ds_test = BaseDS(create_pool_classifiers(), with_IH=True, IH_rate=0.10)
ds_test.fit(X_dsel_ex1, y_dsel_ex1)
ds_test.neighbors = neighbors_ex1[index, :]
ds_test.distances = distances_ex1[index, :]
ds_test.predict_proba_with_ds = MagicMock(return_value=np.atleast_2d([0.25, 0.75]))
proba = ds_test.predict_proba(query)
assert np.allclose(proba, np.atleast_2d([0.25, 0.75]))
def test_predict_proba_IH_knn(index):
query = np.atleast_2d([1, 1])
ds_test = BaseDS(create_pool_classifiers(), with_IH=True, IH_rate=0.5)
ds_test.fit(X_dsel_ex1, y_dsel_ex1)
ds_test.DSEL_scores = dsel_scores_ex1
ds_test.neighbors = neighbors_ex1[index, :]
ds_test.distances = distances_ex1[index, :]
ds_test.roc_algorithm_.predict_proba = MagicMock(return_value=np.atleast_2d([0.45, 0.55]))
proba = ds_test.predict_proba(query)
assert np.allclose(proba, np.atleast_2d([0.45, 0.55]))
def test_label_encoder_only_dsel():
X_dsel_ex1 = np.array([[-1, 1], [-0.75, 0.5], [-1.5, 1.5]])
y_dsel_ex1 = np.array(['cat', 'dog', 'plane'])
query = np.atleast_2d([[1, 0], [-1, -1]])
ds_test = BaseDS(create_pool_classifiers_dog_cat_plane(), k=2)
ds_test.fit(X_dsel_ex1, y_dsel_ex1)
ds_test.classify_with_ds = Mock()
ds_test.classify_with_ds.return_value = [1, 0]
predictions = ds_test.predict(query)
assert np.array_equal(predictions, ['dog', 'cat'])
def test_frienemy_not_all_classifiers_crosses(example_estimate_competence,
create_pool_classifiers):
expected = np.array([[1, 1, 0], [0, 1, 0], [1, 1, 1]])
X, y, neighbors, _, dsel_processed, _ = example_estimate_competence
ds_test = BaseDS(create_pool_classifiers, safe_k=3)
ds_test.fit(X, y)
ds_test.DSEL_processed_ = dsel_processed
# passing three samples to compute the DFP at the same time
result = ds_test._frienemy_pruning(neighbors)
assert np.array_equal(result, expected)
def test_frienemy_safe_region_batch():
n_samples = 10
n_classifiers = 3
expected = np.ones((n_samples, n_classifiers))
ds_test = BaseDS(create_pool_classifiers(), safe_k=3)
ds_test.fit(X_dsel_ex1, y_dsel_ex1)
ds_test.DSEL_processed_ = dsel_processed_ex1
neighbors = np.tile(np.array([0, 1, 2, 6, 7, 8, 14]), (n_samples, 1))
result = ds_test._frienemy_pruning(neighbors)
assert np.array_equal(result, expected)
def test_predict_proba_all_agree(example_estimate_competence,
create_pool_classifiers):
X, y, _, _, _, dsel_scores = example_estimate_competence
query = np.atleast_2d([1, 1])
ds_test = BaseDS(create_pool_classifiers)
ds_test.fit(X, y)
ds_test.DSEL_scores = dsel_scores
backup_all_agree = BaseDS._all_classifier_agree
BaseDS._all_classifier_agree = MagicMock(return_value=np.array([True]))
proba = ds_test.predict_proba(query)
BaseDS._all_classifier_agree = backup_all_agree
assert np.allclose(proba, np.atleast_2d([0.61, 0.39]))
def test_IH_is_used(example_estimate_competence, create_pool_classifiers):
X, y, neighbors, distances, dsel_processed, _ = example_estimate_competence
expected = [0, 0, 1]
query = np.ones((3, 2))
ds_test = BaseDS(create_pool_classifiers, with_IH=True, IH_rate=0.5)
ds_test.fit(X, y)
ds_test.DSEL_processed_ = dsel_processed
ds_test._get_region_competence = MagicMock(return_value=(distances,
neighbors))
predicted = ds_test.predict(query)
assert np.array_equal(predicted, expected)
