def test_no_predict_proba_before_fit(self):
gs = GridSearch(self.estimator, self.disparity_criterion)
X, _, _ = _utilities._quick_data()
with pytest.raises(NotFittedError) as execInfo:
gs.predict_proba(X)
assert _NO_PREDICT_BEFORE_FIT == execInfo.value.args[0]
def test_Y_is_None(self, transformX, transformA, A_two_dim):
gs = GridSearch(self.estimator, self.disparity_criterion)
X, _, A = _utilities._quick_data()
with pytest.raises(ValueError) as execInfo:
gs.fit(transformX(X), None, sensitive_features=transformA(A))
assert _MESSAGE_Y_NONE == execInfo.value.args[0]
def calculate_grid(grid_limit, grid_size, disparity_moment, events):
X, y, A = _utilities._quick_data()
disparity_moment.load_data(X, y, sensitive_features=A)
return _GridGenerator(grid_size, grid_limit,
disparity_moment.pos_basis, disparity_moment.neg_basis,
disparity_moment.neg_basis_present, False).grid
def test_X_is_None(self, transformY, transformA, A_two_dim):
gs = GridSearch(self.estimator, self.disparity_criterion, grid_size=3)
_, Y, A = _utilities._quick_data(A_two_dim)
with pytest.raises(ValueError) as execInfo:
gs.fit(None, transformY(Y), sensitive_features=transformA(A))
assert "Expected 2D array, got scalar array instead" in execInfo.value.args[
0]
def test_Y_not_0_1(self, transformX, transformY, transformA, A_two_dim):
gs = GridSearch(self.estimator, self.disparity_criterion)
X, Y, A = _utilities._quick_data(A_two_dim)
Y = Y + 1
with pytest.raises(ValueError) as execInfo:
gs.fit(transformX(X),
transformY(Y),
sensitive_features=transformA(A))
assert _LABELS_NOT_0_1_ERROR_MESSAGE == execInfo.value.args[0]
def test_Y_ndarray_bad_columns(self, transformX, transformA, A_two_dim):
gs = GridSearch(self.estimator, self.disparity_criterion)
X, Y, A = _utilities._quick_data(A_two_dim)
Y_two_col_ndarray = np.stack((Y, Y), -1)
with pytest.raises(ValueError) as execInfo:
gs.fit(transformX(X),
Y_two_col_ndarray,
sensitive_features=transformA(A))
assert "bad input shape" in execInfo.value.args[0]
def test_Y_df_bad_columns(self, transformX, transformA, A_two_dim):
gs = GridSearch(self.estimator, self.disparity_criterion)
X, Y, A = _utilities._quick_data(A_two_dim)
Y_two_col_df = pd.DataFrame({"a": Y, "b": Y})
with pytest.raises(ValueError) as execInfo:
gs.fit(transformX(X),
Y_two_col_df,
sensitive_features=transformA(A))
assert "bad input shape" in execInfo.value.args[0]
def test_X_Y_different_rows(self, transformX, transformY, transformA,
A_two_dim):
gs = GridSearch(self.estimator, self.disparity_criterion)
X, _, A = _utilities._quick_data()
Y = np.random.randint(2, size=len(A) + 1)
with pytest.raises(ValueError) as execInfo:
gs.fit(transformX(X),
transformY(Y),
sensitive_features=transformA(A))
expected_exception_message = "Found input variables with inconsistent numbers of samples"
assert expected_exception_message in execInfo.value.args[0]
def test_sensitive_feature_non_binary(self, transformX, transformY,
transformA, A_two_dim):
gs = GridSearch(self.estimator, self.disparity_criterion)
X, Y, A = _utilities._quick_data(A_two_dim)
if A_two_dim:
A[0][0] = 0
A[0][1] = 0
A[1][0] = 1
A[1][1] = 1
A[2][0] = 2
A[2][1] = 2
else:
A[0] = 0
A[1] = 1
A[2] = 2
with pytest.raises(ValueError) as execInfo:
gs.fit(transformX(X),
transformY(Y),
sensitive_features=transformA(A))
assert _SENSITIVE_FEATURES_NON_BINARY_ERROR_MESSAGE == execInfo.value.args[
0]
def test_valid_inputs(self, transformX, transformY, transformA, A_two_dim):
gs = GridSearch(self.estimator, self.disparity_criterion, grid_size=2)
X, Y, A = _utilities._quick_data(A_two_dim)
gs.fit(transformX(X), transformY(Y), sensitive_features=transformA(A))
assert len(gs.all_results) == 2
