def test_compute_features_import(self, mock_compute_features_import):
"""
Unit test compute features import
"""
state = SmartState()
state.compute_features_import(Mock())
mock_compute_features_import.assert_called()
def test_compute_grouped_contributions(self, mock_group_contributions):
"""
Unit test compute features groups contributions
"""
state = SmartState()
state.compute_grouped_contributions(Mock(), {})
mock_group_contributions.assert_called()
def test_assign_contributions_2(self):
"""
Unit test assign contributions 2
"""
state = SmartState()
ranked = [1, 2]
with self.assertRaises(ValueError):
state.assign_contributions(ranked)
def test_assign_contributions_1(self):
"""
Unit test assign contributions 1
"""
state = SmartState()
output = state.assign_contributions([1, 2, 3])
expected = {'contrib_sorted': 1, 'x_sorted': 2, 'var_dict': 3}
self.assertDictEqual(output, expected)
def test_check_contributions_1(self):
"""
Unit test check contributions 1
"""
state = SmartState()
contributions = pd.DataFrame([[-0.2, 0.1], [0.8, -0.4], [0.5, -0.7]], )
x_pred = pd.DataFrame([[1, 2], [3, 4]], )
assert not state.check_contributions(contributions, x_pred)
def test_check_contributions_2(self):
"""
Unit test check contributions 2
"""
state = SmartState()
contributions = pd.DataFrame([[-0.2, 0.1], [0.8, -0.4]],
index=['row_1', 'row_2'])
x_pred = pd.DataFrame([[1, 2], [3, 4]])
assert not state.check_contributions(contributions, x_pred)
def test_inverse_transform_contributions(self, mock_inverse_transform_contributions):
"""
Unit test inverse transform contributions
Parameters
----------
mock_inverse_transform_contributions : [type]
[description]
"""
state = SmartState()
state.inverse_transform_contributions(Mock(), Mock())
mock_inverse_transform_contributions.assert_called()
def test_compute_features_import(self, mock_compute_features_import):
"""
Unit test compute features import
Parameters
----------
mock_compute_features_import : [type]
[description]
"""
state = SmartState()
state.compute_features_import(Mock())
mock_compute_features_import.assert_called()
def test_init_mask(self, mock_init_mask):
"""
Unit test init mask
Parameters
----------
mock_init_mask : [type]
[description]
"""
state = SmartState()
state.init_mask(Mock())
mock_init_mask.assert_called()
def test_compute_masked_contributions(self, mock_compute_masked_contributions):
"""
Unit test compute masked contributions
Parameters
----------
mock_compute_masked_contributions : [type]
[description]
"""
state = SmartState()
state.compute_masked_contributions(Mock(), Mock())
mock_compute_masked_contributions.assert_called()
def test_combine_masks(self, mock_combine_masks):
"""
Unit test combine masks
Parameters
----------
mock_combine_masks : [type]
[description]
"""
state = SmartState()
state.combine_masks(Mock())
mock_combine_masks.assert_called()
def test_sign_contributions(self, mock_sign_contributions):
"""
Unit test sign contributions
Parameters
----------
mock_sign_contributions : [type]
[description]
"""
state = SmartState()
state.sign_contributions(Mock(), Mock())
mock_sign_contributions.assert_called()
def test_check_contributions_3(self):
"""
Unit test check contributions 3
"""
state = SmartState()
contributions = pd.DataFrame(
[[-0.2, 0.1], [0.8, -0.4]],
columns=['col_1', 'col_2'],
)
x_pred = pd.DataFrame([[1, 2], [3, 4]], )
assert not state.check_contributions(contributions, x_pred)
def test_validate_contributions_2(self):
"""
Unit test validate contributions 2
"""
state = SmartState()
contributions = np.array([[2, 1], [8, 4]])
x_pred = pd.DataFrame([[1, 2], [3, 4]],
columns=['Col1', 'Col2'],
index=['Id1', 'Id2'])
expected_output = pd.DataFrame([[2, 1], [8, 4]],
columns=['Col1', 'Col2'],
index=['Id1', 'Id2'])
output = state.validate_contributions(contributions, x_pred)
assert not pd.testing.assert_frame_equal(expected_output, output)
def test_summarize_1(self):
"""
Unit test summarize 1
"""
state = SmartState()
contrib_sorted = pd.DataFrame(
[[0.32230754, 0.1550689, 0.10183475, 0.05471339],
[-0.58547512, -0.37050409, -0.07249285, 0.00171975],
[-0.48666675, 0.25507156, -0.16968889, 0.0757443]],
columns=[
'contribution_0', 'contribution_1', 'contribution_2',
'contribution_3'
],
index=[0, 1, 2])
var_dict = pd.DataFrame(
[[1, 0, 2, 3], [1, 0, 3, 2], [1, 0, 2, 3]],
columns=['feature_0', 'feature_1', 'feature_2', 'feature_3'],
index=[0, 1, 2])
x_sorted = pd.DataFrame(
[[1., 3., 22., 1.], [2., 1., 2., 38.], [2., 3., 26., 1.]],
columns=['feature_0', 'feature_1', 'feature_2', 'feature_3'],
index=[0, 1, 2])
mask = pd.DataFrame(
[[True, True, False, False], [True, True, False, False],
[True, True, False, False]],
columns=[
'contribution_0', 'contribution_1', 'contribution_2',
'contribution_3'
],
index=[0, 1, 2])
columns_dict = {0: 'Pclass', 1: 'Sex', 2: 'Age', 3: 'Embarked'}
features_dict = {
'Pclass': 'Pclass',
'Sex': 'Sex',
'Age': 'Age',
'Embarked': 'Embarked'
}
output = state.summarize(contrib_sorted, var_dict, x_sorted, mask,
columns_dict, features_dict)
expected = pd.DataFrame(
[['Sex', 1.0, 0.32230754, 'Pclass', 3.0, 0.1550689],
['Sex', 2.0, -0.58547512, 'Pclass', 1.0, -0.37050409],
['Sex', 2.0, -0.48666675, 'Pclass', 3.0, 0.25507156]],
columns=[
'feature_1', 'value_1', 'contribution_1', 'feature_2',
'value_2', 'contribution_2'
],
index=[0, 1, 2],
dtype=object)
assert not pd.testing.assert_frame_equal(expected, output)
def test_validate_contributions_1(self, choose_state):
"""
Unit test validate contributions 1
"""
choose_state.return_value = MultiDecorator(SmartState())
predictor_1 = self.predictor_1
contributions = [
np.array([[2, 1], [8, 4]]),
np.array([[5, 5], [0, 0]])
]
predictor_1.state = predictor_1.choose_state(contributions)
predictor_1.data = {"x": None, "ypred": None, "contributions": None}
predictor_1.data["x_preprocessed"] = pd.DataFrame(
[[1, 2], [3, 4]], columns=['Col1', 'Col2'], index=['Id1', 'Id2'])
expected_output = [
pd.DataFrame([[2, 1], [8, 4]],
columns=['Col1', 'Col2'],
index=['Id1', 'Id2']),
pd.DataFrame([[5, 5], [0, 0]],
columns=['Col1', 'Col2'],
index=['Id1', 'Id2'])
]
output = predictor_1.validate_contributions(contributions)
assert len(expected_output) == len(output)
test_list = [
pd.testing.assert_frame_equal(e, m)
for e, m in zip(expected_output, output)
]
assert all(x is None for x in test_list)
def test_validate_contributions_1(self):
"""
Unit test validate contributions
Parameters
----------
unittest : [type]
[description]
"""
state = SmartState()
x_pred = Mock()
contributions = pd.DataFrame([[2, 1], [8, 4]],
columns=['Col1', 'Col2'],
index=['Id1', 'Id2'])
expected_output = contributions
output = state.validate_contributions(contributions, x_pred)
assert not pd.testing.assert_frame_equal(expected_output, output)
def test_to_pandas_1(self):
"""
Unit test to pandas 1
"""
xpl = SmartExplainer()
xpl.state = SmartState()
data = {}
data['contrib_sorted'] = pd.DataFrame(
[[0.32230754, 0.1550689, 0.10183475, 0.05471339],
[-0.58547512, -0.37050409, -0.07249285, 0.00171975],
[-0.48666675, 0.25507156, -0.16968889, 0.0757443]],
columns=[
'contribution_0', 'contribution_1', 'contribution_2',
'contribution_3'
],
index=[0, 1, 2])
data['var_dict'] = pd.DataFrame(
[[1, 0, 2, 3], [1, 0, 3, 2], [1, 0, 2, 3]],
columns=['feature_0', 'feature_1', 'feature_2', 'feature_3'],
index=[0, 1, 2])
data['x_sorted'] = pd.DataFrame(
[[1., 3., 22., 1.], [2., 1., 2., 38.], [2., 3., 26., 1.]],
columns=['feature_0', 'feature_1', 'feature_2', 'feature_3'],
index=[0, 1, 2])
xpl.data = data
xpl.columns_dict = {0: 'Pclass', 1: 'Sex', 2: 'Age', 3: 'Embarked'}
xpl.features_dict = {
'Pclass': 'Pclass',
'Sex': 'Sex',
'Age': 'Age',
'Embarked': 'Embarked'
}
xpl.x = pd.DataFrame(
[[3., 1., 22., 1.], [1., 2., 38., 2.], [3., 2., 26., 1.]],
columns=['Pclass', 'Sex', 'Age', 'Embarked'],
index=[0, 1, 2])
xpl.x_pred = xpl.x
xpl.contributions = data['contrib_sorted']
xpl.y_pred = pd.DataFrame([1, 2, 3], columns=['pred'], index=[0, 1, 2])
model = lambda: None
model.predict = types.MethodType(self.predict, model)
xpl.model = model
xpl._case, xpl._classes = xpl.check_model()
xpl.state = xpl.choose_state(xpl.contributions)
output = xpl.to_pandas(max_contrib=2)
expected = pd.DataFrame(
[[1, 'Sex', 1.0, 0.32230754, 'Pclass', 3.0, 0.1550689],
[2, 'Sex', 2.0, -0.58547512, 'Pclass', 1.0, -0.37050409],
[3, 'Sex', 2.0, -0.48666675, 'Pclass', 3.0, 0.25507156]],
columns=[
'pred', 'feature_1', 'value_1', 'contribution_1', 'feature_2',
'value_2', 'contribution_2'
],
index=[0, 1, 2],
dtype=object)
expected['pred'] = expected['pred'].astype(int)
assert not pd.testing.assert_frame_equal(expected, output)
def test_filter_7(self):
"""
Unit test filter 7
"""
xpl = SmartExplainer()
contributions = [
pd.DataFrame(
data=[[0.5, 0.4, 0.3], [0.9, 0.8, 0.7]],
columns=['Col1', 'Col2', 'Col3']
),
pd.DataFrame(
data=[[0.3, 0.2, 0.1], [0.6, 0.5, 0.4]],
columns=['Col1', 'Col2', 'Col3']
)
]
xpl.data = {'var_dict': 1, 'contrib_sorted': contributions, 'x_sorted': 3}
xpl.state = MultiDecorator(SmartState())
xpl.filter(threshold=0.5, max_contrib=2)
expected_mask = [
pd.DataFrame(
data=[[True, False, False], [True, True, False]],
columns=['contrib_1', 'contrib_2', 'contrib_3']
),
pd.DataFrame(
data=[[False, False, False], [True, True, False]],
columns=['contrib_1', 'contrib_2', 'contrib_3']
)
]
assert len(expected_mask) == len(xpl.mask)
test_list = [pd.testing.assert_frame_equal(e, m) for e, m in zip(expected_mask, xpl.mask)]
assert all(x is None for x in test_list)
expected_masked_contributions = [
pd.DataFrame(
data=[[0.0, 0.7], [0.0, 0.7]],
columns=['masked_neg', 'masked_pos']
),
pd.DataFrame(
data=[[0.0, 0.6], [0.0, 0.4]],
columns=['masked_neg', 'masked_pos']
)
]
assert len(expected_masked_contributions) == len(xpl.masked_contributions)
test_list = [pd.testing.assert_frame_equal(e, m) for e, m in
zip(expected_masked_contributions, xpl.masked_contributions)]
assert all(x is None for x in test_list)
expected_param_dict = {
'features_to_hide': None,
'threshold': 0.5,
'positive': None,
'max_contrib': 2
}
self.assertDictEqual(expected_param_dict, xpl.mask_params)
def test_check_contributions(self, choose_state):
"""
Unit test check_shape_contributions 1
"""
choose_state.return_value = MultiDecorator(SmartState())
shap_values = self.clf_2.get_feature_importance(Pool(
self.df_encoded_2),
type="ShapValues")
predictor_1 = self.predictor_2
predictor_1.data = {
"x": None,
"ypred": None,
"contributions": None,
"x_preprocessed": None
}
predictor_1.data["x"] = self.df_2[["x1", "x2"]]
predictor_1.data["x_preprocessed"] = self.df_2[["x1", "x2"]]
predictor_1.data["ypred"] = self.df_2["y"]
adapt_contrib = [
np.array([[-0.04395604, 0.13186813], [-0.04395604, 0.13186813],
[-0.0021978, 0.01318681], [-0.0021978, 0.01318681],
[-0.04395604, 0.13186813]]),
np.array([[0.04395604, -0.13186813], [0.04395604, -0.13186813],
[0.0021978, -0.01318681], [0.0021978, -0.01318681],
[0.04395604, -0.13186813]])
]
contributions = list()
for element in adapt_contrib:
contributions.append(pd.DataFrame(element, columns=["x1", "x2"]))
predictor_1.state = predictor_1.choose_state(adapt_contrib)
predictor_1.check_contributions(contributions)
with self.assertRaises(ValueError):
predictor_1.check_contributions(shap_values[:, :-1])
