def setUp(self):
X_train, y_train, X_test, y_test = titanic_fare()
model = RandomForestRegressor(n_estimators=50,
max_depth=4).fit(X_train, y_train)
X_test.reset_index(drop=True, inplace=True)
X_test.index = X_test.index.astype(str)
X_test1, y_test1 = X_test.iloc[:100], y_test.iloc[:100]
X_test2, y_test2 = X_test.iloc[100:], y_test.iloc[100:]
self.explainer = RegressionExplainer(model,
X_test1,
y_test1,
cats=['Sex', 'Deck'])
def index_exists_func(index):
return index in X_test2.index
def index_list_func():
# only returns first 50 indexes
return list(X_test2.index[:50])
def y_func(index):
idx = X_test2.index.get_loc(index)
return y_test2.iloc[[idx]]
def X_func(index):
idx = X_test2.index.get_loc(index)
return X_test2.iloc[[idx]]
self.explainer.set_index_exists_func(index_exists_func)
self.explainer.set_index_list_func(index_list_func)
self.explainer.set_X_row_func(X_func)
self.explainer.set_y_func(y_func)
def setUp(self):
X_train, y_train, X_test, y_test = titanic_fare()
model = RandomForestRegressor(n_estimators=5, max_depth=2).fit(X_train, y_train)
self.explainer = RegressionExplainer(
model, X_test, y_test,
cats=[{'Gender': ['Sex_female', 'Sex_male', 'Sex_nan']},
'Deck', 'Embarked'],
cv=3)
def setUp(self):
X_train, y_train, X_test, y_test = titanic_fare()
self.test_len = len(X_test)
model = LinearRegression().fit(X_train, y_train)
self.explainer = RegressionExplainer(model,
X_test.iloc[:20],
y_test.iloc[:20],
shap='kernel',
X_background=shap.sample(
X_train, 5))
def setUp(self):
X_train, y_train, X_test, y_test = titanic_fare()
self.test_len = len(X_test)
train_names, test_names = titanic_names()
_, self.names = titanic_names()
model = LGBMRegressor()
model.fit(X_train, y_train)
self.explainer = RegressionExplainer(model, X_test, y_test, r2_score,
shap='tree',
cats=['Sex', 'Deck', 'Embarked'],
idxs=test_names, units="$")
def setUp(self):
X_train, y_train, X_test, y_test = titanic_fare()
self.test_len = len(X_test)
train_names, test_names = titanic_names()
_, self.names = titanic_names()
model = XGBRegressor()
model.fit(X_train, y_train)
self.explainer = RegressionExplainer(model, X_test, y_test,
cats=[{'Gender': ['Sex_female', 'Sex_male', 'Sex_nan']},
'Deck', 'Embarked'],
units="$")
def setUp(self):
X_train, y_train, X_test, y_test = titanic_fare()
self.test_len = len(X_test)
train_names, test_names = titanic_names()
_, self.names = titanic_names()
model = CatBoostRegressor(iterations=5, learning_rate=0.1, verbose=0)
model.fit(X_train, y_train)
self.explainer = RegressionExplainer(model, X_test, y_test,
cats=[{'Gender': ['Sex_female', 'Sex_male', 'Sex_nan']},
'Deck', 'Embarked'],
idxs=test_names, units="$")
def setUp(self):
X_train, y_train, X_test, y_test = titanic_fare()
self.test_len = len(X_test)
train_names, test_names = titanic_names()
_, self.names = titanic_names()
model = RandomForestRegressor(n_estimators=5, max_depth=2).fit(X_train, y_train)
self.explainer = RegressionExplainer(
model, X_test, y_test, r2_score,
cats=[{'Gender': ['Sex_female', 'Sex_male', 'Sex_nan']},
'Deck', 'Embarked'],
idxs=test_names, target='Fare', units='$')
def setUp(self):
X_train, y_train, X_test, y_test = titanic_fare()
self.test_len = len(X_test)
train_names, test_names = titanic_names()
_, self.names = titanic_names()
model = CatBoostRegressor(iterations=100, learning_rate=0.1, verbose=0)
model.fit(X_train, y_train)
self.explainer = RegressionExplainer(model, X_test, y_test, r2_score,
shap='tree',
cats=['Sex', 'Deck', 'Embarked'],
idxs=test_names, units="$")
class CatBoostRegressionTests(unittest.TestCase):
def setUp(self):
X_train, y_train, X_test, y_test = titanic_fare()
self.test_len = len(X_test)
train_names, test_names = titanic_names()
_, self.names = titanic_names()
model = CatBoostRegressor(iterations=100, learning_rate=0.1, verbose=0)
model.fit(X_train, y_train)
self.explainer = RegressionExplainer(model, X_test, y_test, r2_score,
shap='tree',
cats=['Sex', 'Deck', 'Embarked'],
idxs=test_names, units="$")
def test_preds(self):
self.assertIsInstance(self.explainer.preds, np.ndarray)
def test_permutation_importances(self):
self.assertIsInstance(self.explainer.permutation_importances, pd.DataFrame)
self.assertIsInstance(self.explainer.permutation_importances_cats, pd.DataFrame)
def test_shap_base_value(self):
self.assertIsInstance(self.explainer.shap_base_value, (np.floating, float))
def test_shap_values_shape(self):
self.assertTrue(self.explainer.shap_values.shape == (len(self.explainer), len(self.explainer.columns)))
def test_shap_values(self):
self.assertIsInstance(self.explainer.shap_values, np.ndarray)
self.assertIsInstance(self.explainer.shap_values_cats, np.ndarray)
# @unittest.expectedFailure
# def test_shap_interaction_values(self):
# self.assertIsInstance(self.explainer.shap_interaction_values, np.ndarray)
# self.assertIsInstance(self.explainer.shap_interaction_values_cats, np.ndarray)
def test_mean_abs_shap(self):
self.assertIsInstance(self.explainer.mean_abs_shap, pd.DataFrame)
self.assertIsInstance(self.explainer.mean_abs_shap_cats, pd.DataFrame)
def test_calculate_properties(self):
self.explainer.calculate_properties(include_interactions=False)
def test_pdp_result(self):
self.assertIsInstance(self.explainer.get_pdp_result("Age"), pdpbox.pdp.PDPIsolate)
self.assertIsInstance(self.explainer.get_pdp_result("Sex"), pdpbox.pdp.PDPIsolate)
self.assertIsInstance(self.explainer.get_pdp_result("Age", index=0), pdpbox.pdp.PDPIsolate)
self.assertIsInstance(self.explainer.get_pdp_result("Sex", index=0), pdpbox.pdp.PDPIsolate)
class SkorchRegressorTests(unittest.TestCase):
def setUp(self):
model, X, y = get_skorch_regressor()
self.explainer = RegressionExplainer(model, X, y)
def test_preds(self):
self.assertIsInstance(self.explainer.preds, np.ndarray)
def test_permutation_importances(self):
self.assertIsInstance(self.explainer.get_permutation_importances_df(),
pd.DataFrame)
def test_shap_base_value(self):
self.assertIsInstance(self.explainer.shap_base_value(),
(np.floating, float))
def test_shap_values_shape(self):
self.assertTrue(self.explainer.get_shap_values_df().shape == (
len(self.explainer), len(self.explainer.merged_cols)))
def test_shap_values(self):
self.assertIsInstance(self.explainer.get_shap_values_df(),
pd.DataFrame)
def test_mean_abs_shap(self):
self.assertIsInstance(self.explainer.get_mean_abs_shap_df(),
pd.DataFrame)
def test_calculate_properties(self):
self.explainer.calculate_properties(include_interactions=False)
def test_pdp_df(self):
self.assertIsInstance(self.explainer.pdp_df("col1"), pd.DataFrame)
class RegressionCVTests(unittest.TestCase):
def setUp(self):
X_train, y_train, X_test, y_test = titanic_fare()
model = RandomForestRegressor(n_estimators=5, max_depth=2).fit(X_train, y_train)
self.explainer = RegressionExplainer(
model, X_test, y_test,
cats=[{'Gender': ['Sex_female', 'Sex_male', 'Sex_nan']},
'Deck', 'Embarked'],
cv=3)
def test_cv_permutation_importances(self):
self.assertIsInstance(self.explainer.permutation_importances(), pd.DataFrame)
def test_cv_metrics(self):
self.assertIsInstance(self.explainer.metrics(), dict)
def setUp(self):
X_train, y_train, X_test, y_test = titanic_fare()
self.test_len = len(X_test)
train_names, test_names = titanic_names()
_, self.names = titanic_names()
model = RandomForestRegressor(n_estimators=5, max_depth=2)
model.fit(X_train, y_train)
self.explainer = RegressionExplainer(model,
X_test,
y_test,
r2_score,
shap='tree',
cats=['Sex', 'Cabin', 'Embarked'],
idxs=test_names)
def setUp(self):
X_train, y_train, X_test, y_test = titanic_fare()
self.test_len = len(X_test)
train_names, test_names = titanic_names()
_, self.names = titanic_names()
model = RandomForestRegressor(n_estimators=5, max_depth=2)
model.fit(X_train, y_train)
self.explainer = RegressionExplainer(
model,
X_test,
y_test,
cats=[{
'Gender': ['Sex_female', 'Sex_male', 'Sex_nan']
}, 'Deck', 'Embarked'],
cats_notencoded={'Gender': 'No Gender'},
idxs=test_names)
def setUp(self):
X_train, y_train, X_test, y_test = titanic_fare()
self.test_len = len(X_test)
train_names, test_names = titanic_names()
_, self.names = titanic_names()
model = LinearRegression()
model.fit(X_train, y_train)
self.explainer = RegressionExplainer(
model,
X_test,
y_test,
r2_score,
shap='linear',
cats=[{
'Gender': ['Sex_female', 'Sex_male', 'Sex_nan']
}, 'Deck', 'Embarked'],
idxs=test_names,
units="$")
class LinearRegressionKernelTests(unittest.TestCase):
def setUp(self):
X_train, y_train, X_test, y_test = titanic_fare()
self.test_len = len(X_test)
model = LinearRegression().fit(X_train, y_train)
self.explainer = RegressionExplainer(model,
X_test.iloc[:20],
y_test.iloc[:20],
shap='kernel',
X_background=shap.sample(
X_train, 5))
def test_shap_values(self):
self.assertIsInstance(self.explainer.shap_base_value(),
(np.floating, float))
self.assertTrue(self.explainer.get_shap_values_df().shape == (
len(self.explainer), len(self.explainer.merged_cols)))
self.assertIsInstance(self.explainer.get_shap_values_df(),
pd.DataFrame)
def get_regression_explainer(xgboost=False, include_y=True):
X_train, y_train, X_test, y_test = titanic_fare()
train_names, test_names = titanic_names()
if xgboost:
model = XGBRegressor().fit(X_train, y_train)
else:
model = RandomForestRegressor(n_estimators=50,
max_depth=10).fit(X_train, y_train)
if include_y:
reg_explainer = RegressionExplainer(model,
X_test,
y_test,
cats=['Sex', 'Deck', 'Embarked'],
idxs=test_names,
units="$")
else:
reg_explainer = RegressionExplainer(model,
X_test,
cats=['Sex', 'Deck', 'Embarked'],
idxs=test_names,
units="$")
reg_explainer.calculate_properties()
return reg_explainer
def setUp(self):
X_train, y_train, X_test, y_test = titanic_fare()
train_names, test_names = titanic_names()
model = XGBRegressor(n_estimators=5, max_depth=2)
model.fit(X_train, y_train)
self.explainer = RegressionExplainer(
model,
X_test,
y_test,
cats=[{
'Gender': ['Sex_female', 'Sex_male', 'Sex_nan']
}, 'Deck', 'Embarked'],
idxs=test_names)
def get_catboost_regressor():
X_train, y_train, X_test, y_test = titanic_fare()
model = CatBoostRegressor(iterations=5, verbose=0).fit(X_train, y_train)
explainer = RegressionExplainer(model, X_test, y_test,
cats=["Sex", 'Deck', 'Embarked'])
X_cats, y_cats = explainer.X_merged, explainer.y
model = CatBoostRegressor(iterations=5, verbose=0).fit(X_cats, y_cats, cat_features=[5, 6, 7])
explainer = RegressionExplainer(model, X_cats, y_cats, idxs=X_test.index)
explainer.calculate_properties(include_interactions=False)
return explainer
class XGBRegressionExplainerTests(unittest.TestCase):
def setUp(self):
X_train, y_train, X_test, y_test = titanic_fare()
self.test_len = len(X_test)
train_names, test_names = titanic_names()
_, self.names = titanic_names()
model = XGBRegressor(n_estimators=5, max_depth=2)
model.fit(X_train, y_train)
self.explainer = RegressionExplainer(
model,
X_test,
y_test,
cats=[{
'Gender': ['Sex_female', 'Sex_male', 'Sex_nan']
}, 'Deck', 'Embarked'],
idxs=test_names)
def test_graphviz_available(self):
self.assertIsInstance(self.explainer.graphviz_available, bool)
def test_shadow_trees(self):
dt = self.explainer.shadow_trees
self.assertIsInstance(dt, list)
self.assertIsInstance(
dt[0], dtreeviz.models.shadow_decision_tree.ShadowDecTree)
def test_decisionpath_df(self):
df = self.explainer.get_decisionpath_df(tree_idx=0, index=0)
self.assertIsInstance(df, pd.DataFrame)
df = self.explainer.get_decisionpath_df(tree_idx=0,
index=self.names[0])
self.assertIsInstance(df, pd.DataFrame)
def test_plot_trees(self):
fig = self.explainer.plot_trees(index=0)
self.assertIsInstance(fig, go.Figure)
fig = self.explainer.plot_trees(index=self.names[0])
self.assertIsInstance(fig, go.Figure)
fig = self.explainer.plot_trees(index=self.names[0], highlight_tree=0)
self.assertIsInstance(fig, go.Figure)
def test_calculate_properties(self):
self.explainer.calculate_properties()
class LGBMRegressionTests(unittest.TestCase):
def setUp(self):
X_train, y_train, X_test, y_test = titanic_fare()
self.test_len = len(X_test)
train_names, test_names = titanic_names()
_, self.names = titanic_names()
model = LGBMRegressor()
model.fit(X_train, y_train)
self.explainer = RegressionExplainer(model, X_test, y_test, r2_score,
shap='tree',
cats=[{'Gender': ['Sex_female', 'Sex_male', 'Sex_nan']},
'Deck', 'Embarked'],
idxs=test_names, units="$")
def test_preds(self):
self.assertIsInstance(self.explainer.preds, np.ndarray)
def test_permutation_importances(self):
self.assertIsInstance(self.explainer.permutation_importances, pd.DataFrame)
self.assertIsInstance(self.explainer.permutation_importances_cats, pd.DataFrame)
def test_shap_base_value(self):
self.assertIsInstance(self.explainer.shap_base_value, (np.floating, float))
def test_shap_values_shape(self):
self.assertTrue(self.explainer.shap_values.shape == (len(self.explainer), len(self.explainer.columns)))
def test_shap_values(self):
self.assertIsInstance(self.explainer.shap_values, np.ndarray)
self.assertIsInstance(self.explainer.shap_values_cats, np.ndarray)
# def test_shap_interaction_values(self):
# self.assertIsInstance(self.explainer.shap_interaction_values, np.ndarray)
# self.assertIsInstance(self.explainer.shap_interaction_values_cats, np.ndarray)
def test_mean_abs_shap(self):
self.assertIsInstance(self.explainer.mean_abs_shap, pd.DataFrame)
self.assertIsInstance(self.explainer.mean_abs_shap_cats, pd.DataFrame)
def test_calculate_properties(self):
self.explainer.calculate_properties(include_interactions=True)
def test_pdp_result(self):
self.assertIsInstance(self.explainer.get_pdp_result("Age"), pdpbox.pdp.PDPIsolate)
self.assertIsInstance(self.explainer.get_pdp_result("Gender"), pdpbox.pdp.PDPIsolate)
self.assertIsInstance(self.explainer.get_pdp_result("Age", index=0), pdpbox.pdp.PDPIsolate)
self.assertIsInstance(self.explainer.get_pdp_result("Gender", index=0), pdpbox.pdp.PDPIsolate)
class RegressionBaseExplainerTests(unittest.TestCase):
def setUp(self):
X_train, y_train, X_test, y_test = titanic_fare()
self.test_len = len(X_test)
train_names, test_names = titanic_names()
_, self.names = titanic_names()
model = RandomForestRegressor(n_estimators=5, max_depth=2).fit(X_train, y_train)
self.explainer = RegressionExplainer(
model, X_test, y_test, r2_score,
cats=[{'Gender': ['Sex_female', 'Sex_male', 'Sex_nan']},
'Deck', 'Embarked'],
idxs=test_names, target='Fare', units='$')
def test_explainer_len(self):
self.assertEqual(len(self.explainer), self.test_len)
def test_int_idx(self):
self.assertEqual(self.explainer.get_idx(self.names[0]), 0)
def test_random_index(self):
self.assertIsInstance(self.explainer.random_index(), int)
self.assertIsInstance(self.explainer.random_index(return_str=True), str)
def test_row_from_input(self):
input_row = self.explainer.get_row_from_input(
self.explainer.X.iloc[[0]].values.tolist())
self.assertIsInstance(input_row, pd.DataFrame)
input_row = self.explainer.get_row_from_input(
self.explainer.X_merged.iloc[[0]].values.tolist())
self.assertIsInstance(input_row, pd.DataFrame)
input_row = self.explainer.get_row_from_input(
self.explainer.X_merged
[self.explainer.columns_ranked_by_shap()]
.iloc[[0]].values.tolist(), ranked_by_shap=True)
self.assertIsInstance(input_row, pd.DataFrame)
def test_prediction_result_df(self):
df = self.explainer.prediction_result_df(0)
self.assertIsInstance(df, pd.DataFrame)
def test_preds(self):
self.assertIsInstance(self.explainer.preds, np.ndarray)
def test_pred_percentiles(self):
self.assertIsInstance(self.explainer.pred_percentiles(), np.ndarray)
def test_columns_ranked_by_shap(self):
self.assertIsInstance(self.explainer.columns_ranked_by_shap(), list)
def test_get_col(self):
self.assertIsInstance(self.explainer.get_col("Gender"), pd.Series)
self.assertTrue(is_categorical_dtype(self.explainer.get_col("Gender")))
self.assertIsInstance(self.explainer.get_col("Age"), pd.Series)
self.assertTrue(is_numeric_dtype(self.explainer.get_col("Age")))
def test_permutation_importances(self):
self.assertIsInstance(self.explainer.permutation_importances(), pd.DataFrame)
def test_X_cats(self):
self.assertIsInstance(self.explainer.X_cats, pd.DataFrame)
def test_metrics(self):
self.assertIsInstance(self.explainer.metrics(), dict)
self.assertIsInstance(self.explainer.metrics_descriptions(), dict)
def test_mean_abs_shap_df(self):
self.assertIsInstance(self.explainer.mean_abs_shap_df(), pd.DataFrame)
def test_top_interactions(self):
self.assertIsInstance(self.explainer.top_shap_interactions("Age"), list)
self.assertIsInstance(self.explainer.top_shap_interactions("Age", topx=4), list)
def test_permutation_importances_df(self):
self.assertIsInstance(self.explainer.get_permutation_importances_df(), pd.DataFrame)
self.assertIsInstance(self.explainer.get_permutation_importances_df(topx=3), pd.DataFrame)
self.assertIsInstance(self.explainer.get_permutation_importances_df(cutoff=0.01), pd.DataFrame)
def test_contrib_df(self):
self.assertIsInstance(self.explainer.get_contrib_df(0), pd.DataFrame)
self.assertIsInstance(self.explainer.get_contrib_df(0, topx=3), pd.DataFrame)
self.assertIsInstance(self.explainer.get_contrib_df(0, sort='high-to-low'), pd.DataFrame)
self.assertIsInstance(self.explainer.get_contrib_df(0, sort='low-to-high'), pd.DataFrame)
self.assertIsInstance(self.explainer.get_contrib_df(0, sort='importance'), pd.DataFrame)
self.assertIsInstance(self.explainer.get_contrib_df(X_row=self.explainer.X.iloc[[0]]), pd.DataFrame)
def test_contrib_summary_df(self):
self.assertIsInstance(self.explainer.get_contrib_summary_df(0), pd.DataFrame)
self.assertIsInstance(self.explainer.get_contrib_summary_df(0, topx=3), pd.DataFrame)
self.assertIsInstance(self.explainer.get_contrib_summary_df(0, round=3), pd.DataFrame)
self.assertIsInstance(self.explainer.get_contrib_summary_df(0, sort='high-to-low'), pd.DataFrame)
self.assertIsInstance(self.explainer.get_contrib_summary_df(0, sort='low-to-high'), pd.DataFrame)
self.assertIsInstance(self.explainer.get_contrib_summary_df(0, sort='importance'), pd.DataFrame)
self.assertIsInstance(self.explainer.get_contrib_summary_df(X_row=self.explainer.X.iloc[[0]]), pd.DataFrame)
def test_shap_base_value(self):
self.assertIsInstance(self.explainer.shap_base_value(), (np.floating, float))
def test_shap_values_shape(self):
self.assertTrue(self.explainer.get_shap_values_df().shape == (len(self.explainer), len(self.explainer.merged_cols)))
def test_shap_values(self):
self.assertIsInstance(self.explainer.get_shap_values_df(), pd.DataFrame)
def test_shap_interaction_values(self):
self.assertIsInstance(self.explainer.shap_interaction_values(), np.ndarray)
def test_mean_abs_shap(self):
self.assertIsInstance(self.explainer.mean_abs_shap_df(), pd.DataFrame)
def test_memory_usage(self):
self.assertIsInstance(self.explainer.memory_usage(), pd.DataFrame)
self.assertIsInstance(self.explainer.memory_usage(cutoff=1000), pd.DataFrame)
def test_calculate_properties(self):
self.explainer.calculate_properties()
def test_shap_interaction_values_for_col(self):
self.assertIsInstance(self.explainer.shap_interaction_values_for_col("Age"), np.ndarray)
self.assertEqual(self.explainer.shap_interaction_values_for_col("Age").shape,
self.explainer.get_shap_values_df().shape)
def test_pdp_df(self):
self.assertIsInstance(self.explainer.pdp_df("Age"), pd.DataFrame)
self.assertIsInstance(self.explainer.pdp_df("Gender"), pd.DataFrame)
self.assertIsInstance(self.explainer.pdp_df("Deck"), pd.DataFrame)
self.assertIsInstance(self.explainer.pdp_df("Age", index=0), pd.DataFrame)
self.assertIsInstance(self.explainer.pdp_df("Gender", index=0), pd.DataFrame)
def test_plot_importances(self):
fig = self.explainer.plot_importances()
self.assertIsInstance(fig, go.Figure)
fig = self.explainer.plot_importances(kind='permutation')
self.assertIsInstance(fig, go.Figure)
fig = self.explainer.plot_importances(topx=3)
self.assertIsInstance(fig, go.Figure)
def test_plot_interactions(self):
fig = self.explainer.plot_interactions_importance("Age")
self.assertIsInstance(fig, go.Figure)
fig = self.explainer.plot_interactions_importance("Age")
self.assertIsInstance(fig, go.Figure)
fig = self.explainer.plot_interactions_importance("Gender")
self.assertIsInstance(fig, go.Figure)
def test_plot_shap_interactions(self):
fig = self.explainer.plot_contributions(0)
self.assertIsInstance(fig, go.Figure)
fig = self.explainer.plot_contributions(0, topx=3)
self.assertIsInstance(fig, go.Figure)
fig = self.explainer.plot_contributions(0, cutoff=0.05)
self.assertIsInstance(fig, go.Figure)
fig = self.explainer.plot_contributions(0, sort='high-to-low')
self.assertIsInstance(fig, go.Figure)
fig = self.explainer.plot_contributions(0, sort='low-to-high')
self.assertIsInstance(fig, go.Figure)
fig = self.explainer.plot_contributions(0, sort='importance')
self.assertIsInstance(fig, go.Figure)
def test_plot_shap_detailed(self):
fig = self.explainer.plot_importances_detailed()
self.assertIsInstance(fig, go.Figure)
fig = self.explainer.plot_importances_detailed(topx=3)
self.assertIsInstance(fig, go.Figure)
def test_plot_interactions_detailed(self):
fig = self.explainer.plot_interactions_detailed("Age")
self.assertIsInstance(fig, go.Figure)
fig = self.explainer.plot_interactions_detailed("Age", topx=3)
self.assertIsInstance(fig, go.Figure)
def test_plot_dependence(self):
fig = self.explainer.plot_dependence("Age")
self.assertIsInstance(fig, go.Figure)
fig = self.explainer.plot_dependence("Gender")
self.assertIsInstance(fig, go.Figure)
fig = self.explainer.plot_dependence("Age", "Gender")
self.assertIsInstance(fig, go.Figure)
fig = self.explainer.plot_dependence("Age", highlight_index=0)
self.assertIsInstance(fig, go.Figure)
fig = self.explainer.plot_dependence("Gender", highlight_index=0)
self.assertIsInstance(fig, go.Figure)
def test_plot_contributions(self):
fig = self.explainer.plot_contributions(0)
self.assertIsInstance(fig, go.Figure)
fig = self.explainer.plot_contributions(0, topx=3)
self.assertIsInstance(fig, go.Figure)
fig = self.explainer.plot_contributions(0, sort='high-to-low')
self.assertIsInstance(fig, go.Figure)
fig = self.explainer.plot_contributions(0, sort='low-to-high')
self.assertIsInstance(fig, go.Figure)
fig = self.explainer.plot_contributions(0, sort='importance')
self.assertIsInstance(fig, go.Figure)
fig = self.explainer.plot_contributions(X_row=self.explainer.X.iloc[[0]])
self.assertIsInstance(fig, go.Figure)
def test_plot_interaction(self):
fig = self.explainer.plot_interaction("Age", "Gender")
self.assertIsInstance(fig, go.Figure)
fig = self.explainer.plot_interaction("Gender", "Age")
self.assertIsInstance(fig, go.Figure)
fig = self.explainer.plot_interaction("Gender", "Age")
self.assertIsInstance(fig, go.Figure)
fig = self.explainer.plot_interaction("Age", "Gender", highlight_index=0)
self.assertIsInstance(fig, go.Figure)
def test_plot_pdp(self):
fig = self.explainer.plot_pdp("Age")
self.assertIsInstance(fig, go.Figure)
fig = self.explainer.plot_pdp("Gender")
self.assertIsInstance(fig, go.Figure)
fig = self.explainer.plot_pdp("Gender", index=0)
self.assertIsInstance(fig, go.Figure)
fig = self.explainer.plot_pdp("Age", index=0)
self.assertIsInstance(fig, go.Figure)
fig = self.explainer.plot_pdp("Age", X_row=self.explainer.X.iloc[[0]])
self.assertIsInstance(fig, go.Figure)
def test_yaml(self):
yaml = self.explainer.to_yaml()
self.assertIsInstance(yaml, str)
class RegressionBaseExplainerTests(unittest.TestCase):
def setUp(self):
X_train, y_train, X_test, y_test = titanic_fare()
self.test_len = len(X_test)
train_names, test_names = titanic_names()
_, self.names = titanic_names()
model = RandomForestRegressor(n_estimators=5, max_depth=2)
model.fit(X_train, y_train)
self.explainer = RegressionExplainer(
model,
X_test,
y_test,
r2_score,
cats=[{
'Gender': ['Sex_female', 'Sex_male', 'Sex_nan']
}, 'Deck', 'Embarked'],
idxs=test_names,
target='Fare',
units='$')
def test_explainer_len(self):
self.assertEqual(len(self.explainer), self.test_len)
def test_int_idx(self):
self.assertEqual(self.explainer.get_int_idx(self.names[0]), 0)
def test_random_index(self):
self.assertIsInstance(self.explainer.random_index(), int)
self.assertIsInstance(self.explainer.random_index(return_str=True),
str)
def test_preds(self):
self.assertIsInstance(self.explainer.preds, np.ndarray)
def test_pred_percentiles(self):
self.assertIsInstance(self.explainer.pred_percentiles, np.ndarray)
def test_columns_ranked_by_shap(self):
self.assertIsInstance(self.explainer.columns_ranked_by_shap(), list)
self.assertIsInstance(self.explainer.columns_ranked_by_shap(cats=True),
list)
def test_equivalent_col(self):
self.assertEqual(self.explainer.equivalent_col("Sex_female"), "Gender")
self.assertEqual(self.explainer.equivalent_col("Gender"), "Sex_female")
self.assertIsNone(self.explainer.equivalent_col("random"))
def test_get_col(self):
self.assertIsInstance(self.explainer.get_col("Gender"), pd.Series)
self.assertEqual(self.explainer.get_col("Gender").dtype, "object")
self.assertIsInstance(self.explainer.get_col("Age"), pd.Series)
self.assertEqual(self.explainer.get_col("Age").dtype, np.float)
def test_permutation_importances(self):
self.assertIsInstance(self.explainer.permutation_importances,
pd.DataFrame)
self.assertIsInstance(self.explainer.permutation_importances_cats,
pd.DataFrame)
def test_X_cats(self):
self.assertIsInstance(self.explainer.X_cats, pd.DataFrame)
def test_columns_cats(self):
self.assertIsInstance(self.explainer.columns_cats, list)
def test_metrics(self):
self.assertIsInstance(self.explainer.metrics(), dict)
self.assertIsInstance(self.explainer.metrics_markdown(), str)
def test_mean_abs_shap_df(self):
self.assertIsInstance(self.explainer.mean_abs_shap_df(), pd.DataFrame)
def test_top_interactions(self):
self.assertIsInstance(self.explainer.shap_top_interactions("Age"),
list)
self.assertIsInstance(
self.explainer.shap_top_interactions("Age", topx=4), list)
self.assertIsInstance(
self.explainer.shap_top_interactions("Age", cats=True), list)
self.assertIsInstance(
self.explainer.shap_top_interactions("Gender", cats=True), list)
def test_permutation_importances_df(self):
self.assertIsInstance(self.explainer.permutation_importances_df(),
pd.DataFrame)
self.assertIsInstance(
self.explainer.permutation_importances_df(topx=3), pd.DataFrame)
self.assertIsInstance(
self.explainer.permutation_importances_df(cats=True), pd.DataFrame)
self.assertIsInstance(
self.explainer.permutation_importances_df(cutoff=0.01),
pd.DataFrame)
def test_contrib_df(self):
self.assertIsInstance(self.explainer.contrib_df(0), pd.DataFrame)
self.assertIsInstance(self.explainer.contrib_df(0, cats=False),
pd.DataFrame)
self.assertIsInstance(self.explainer.contrib_df(0, topx=3),
pd.DataFrame)
self.assertIsInstance(self.explainer.contrib_df(0, sort='high-to-low'),
pd.DataFrame)
self.assertIsInstance(self.explainer.contrib_df(0, sort='low-to-high'),
pd.DataFrame)
self.assertIsInstance(self.explainer.contrib_df(0, sort='importance'),
pd.DataFrame)
self.assertIsInstance(
self.explainer.contrib_df(X_row=self.explainer.X.iloc[[0]]),
pd.DataFrame)
self.assertIsInstance(
self.explainer.contrib_df(X_row=self.explainer.X_cats.iloc[[0]]),
pd.DataFrame)
def test_contrib_summary_df(self):
self.assertIsInstance(self.explainer.contrib_summary_df(0),
pd.DataFrame)
self.assertIsInstance(self.explainer.contrib_summary_df(0, cats=False),
pd.DataFrame)
self.assertIsInstance(self.explainer.contrib_summary_df(0, topx=3),
pd.DataFrame)
self.assertIsInstance(self.explainer.contrib_summary_df(0, round=3),
pd.DataFrame)
self.assertIsInstance(
self.explainer.contrib_summary_df(0, sort='high-to-low'),
pd.DataFrame)
self.assertIsInstance(
self.explainer.contrib_summary_df(0, sort='low-to-high'),
pd.DataFrame)
self.assertIsInstance(
self.explainer.contrib_summary_df(0, sort='importance'),
pd.DataFrame)
self.assertIsInstance(
self.explainer.contrib_summary_df(
X_row=self.explainer.X.iloc[[0]]), pd.DataFrame)
self.assertIsInstance(
self.explainer.contrib_summary_df(
X_row=self.explainer.X_cats.iloc[[0]]), pd.DataFrame)
def test_shap_base_value(self):
self.assertIsInstance(self.explainer.shap_base_value,
(np.floating, float))
def test_shap_values_shape(self):
self.assertTrue(
self.explainer.shap_values.shape == (len(self.explainer),
len(self.explainer.columns)))
def test_shap_values(self):
self.assertIsInstance(self.explainer.shap_values, np.ndarray)
self.assertIsInstance(self.explainer.shap_values_cats, np.ndarray)
def test_shap_interaction_values(self):
self.assertIsInstance(self.explainer.shap_interaction_values,
np.ndarray)
self.assertIsInstance(self.explainer.shap_interaction_values_cats,
np.ndarray)
def test_mean_abs_shap(self):
self.assertIsInstance(self.explainer.mean_abs_shap, pd.DataFrame)
self.assertIsInstance(self.explainer.mean_abs_shap_cats, pd.DataFrame)
def test_calculate_properties(self):
self.explainer.calculate_properties()
def test_shap_interaction_values_by_col(self):
self.assertIsInstance(
self.explainer.shap_interaction_values_by_col("Age"), np.ndarray)
self.assertEquals(
self.explainer.shap_interaction_values_by_col("Age").shape,
self.explainer.shap_values.shape)
self.assertEquals(
self.explainer.shap_interaction_values_by_col("Age",
cats=True).shape,
self.explainer.shap_values_cats.shape)
def test_pdp_result(self):
self.assertIsInstance(self.explainer.get_pdp_result("Age"),
pdpbox.pdp.PDPIsolate)
self.assertIsInstance(self.explainer.get_pdp_result("Gender"),
pdpbox.pdp.PDPIsolate)
self.assertIsInstance(self.explainer.get_pdp_result("Age", index=0),
pdpbox.pdp.PDPIsolate)
self.assertIsInstance(self.explainer.get_pdp_result("Gender", index=0),
pdpbox.pdp.PDPIsolate)
self.assertIsInstance(
self.explainer.get_pdp_result("Age",
X_row=self.explainer.X.iloc[[0]]),
pdpbox.pdp.PDPIsolate)
self.assertIsInstance(
self.explainer.get_pdp_result(
"Gender", X_row=self.explainer.X_cats.iloc[[0]]),
pdpbox.pdp.PDPIsolate)
def test_get_dfs(self):
cols_df, shap_df, contribs_df = self.explainer.get_dfs()
self.assertIsInstance(cols_df, pd.DataFrame)
self.assertIsInstance(shap_df, pd.DataFrame)
self.assertIsInstance(contribs_df, pd.DataFrame)
def test_plot_importances(self):
fig = self.explainer.plot_importances()
self.assertIsInstance(fig, go.Figure)
fig = self.explainer.plot_importances(kind='permutation')
self.assertIsInstance(fig, go.Figure)
fig = self.explainer.plot_importances(topx=3)
self.assertIsInstance(fig, go.Figure)
fig = self.explainer.plot_importances(cats=True)
self.assertIsInstance(fig, go.Figure)
def test_plot_interactions(self):
fig = self.explainer.plot_interactions("Age")
self.assertIsInstance(fig, go.Figure)
fig = self.explainer.plot_interactions("Sex_female")
self.assertIsInstance(fig, go.Figure)
fig = self.explainer.plot_interactions("Age")
self.assertIsInstance(fig, go.Figure)
fig = self.explainer.plot_interactions("Gender")
self.assertIsInstance(fig, go.Figure)
def test_plot_shap_interactions(self):
fig = self.explainer.plot_shap_contributions(0)
self.assertIsInstance(fig, go.Figure)
fig = self.explainer.plot_shap_contributions(0, cats=False)
self.assertIsInstance(fig, go.Figure)
fig = self.explainer.plot_shap_contributions(0, topx=3)
self.assertIsInstance(fig, go.Figure)
fig = self.explainer.plot_shap_contributions(0, cutoff=0.05)
self.assertIsInstance(fig, go.Figure)
fig = self.explainer.plot_shap_contributions(0, sort='high-to-low')
self.assertIsInstance(fig, go.Figure)
fig = self.explainer.plot_shap_contributions(0, sort='low-to-high')
self.assertIsInstance(fig, go.Figure)
fig = self.explainer.plot_shap_contributions(0, sort='importance')
self.assertIsInstance(fig, go.Figure)
def test_plot_shap_summary(self):
fig = self.explainer.plot_shap_summary()
self.assertIsInstance(fig, go.Figure)
fig = self.explainer.plot_shap_summary(topx=3)
self.assertIsInstance(fig, go.Figure)
fig = self.explainer.plot_shap_summary(cats=True)
self.assertIsInstance(fig, go.Figure)
def test_plot_shap_interaction_summary(self):
fig = self.explainer.plot_shap_interaction_summary("Age")
self.assertIsInstance(fig, go.Figure)
fig = self.explainer.plot_shap_interaction_summary("Age", topx=3)
self.assertIsInstance(fig, go.Figure)
fig = self.explainer.plot_shap_interaction_summary("Age", cats=True)
self.assertIsInstance(fig, go.Figure)
fig = self.explainer.plot_shap_interaction_summary("Sex_female",
topx=3)
self.assertIsInstance(fig, go.Figure)
fig = self.explainer.plot_shap_interaction_summary("Gender", cats=True)
self.assertIsInstance(fig, go.Figure)
def test_plot_shap_dependence(self):
fig = self.explainer.plot_shap_dependence("Age")
self.assertIsInstance(fig, go.Figure)
fig = self.explainer.plot_shap_dependence("Gender")
self.assertIsInstance(fig, go.Figure)
fig = self.explainer.plot_shap_dependence("Age", "Gender")
self.assertIsInstance(fig, go.Figure)
fig = self.explainer.plot_shap_dependence("Sex_female", "Age")
self.assertIsInstance(fig, go.Figure)
fig = self.explainer.plot_shap_dependence("Age", highlight_index=0)
self.assertIsInstance(fig, go.Figure)
fig = self.explainer.plot_shap_dependence("Gender", highlight_index=0)
self.assertIsInstance(fig, go.Figure)
def test_plot_shap_contributions(self):
fig = self.explainer.plot_shap_contributions(0)
self.assertIsInstance(fig, go.Figure)
fig = self.explainer.plot_shap_contributions(0, cats=False)
self.assertIsInstance(fig, go.Figure)
fig = self.explainer.plot_shap_contributions(0, topx=3)
self.assertIsInstance(fig, go.Figure)
fig = self.explainer.plot_shap_contributions(0, sort='high-to-low')
self.assertIsInstance(fig, go.Figure)
fig = self.explainer.plot_shap_contributions(0, sort='low-to-high')
self.assertIsInstance(fig, go.Figure)
fig = self.explainer.plot_shap_contributions(0, sort='importance')
self.assertIsInstance(fig, go.Figure)
fig = self.explainer.plot_shap_contributions(
X_row=self.explainer.X.iloc[[0]])
self.assertIsInstance(fig, go.Figure)
fig = self.explainer.plot_shap_contributions(
X_row=self.explainer.X_cats.iloc[[0]])
self.assertIsInstance(fig, go.Figure)
def test_plot_shap_interaction(self):
fig = self.explainer.plot_shap_interaction("Age", "Sex_female")
self.assertIsInstance(fig, go.Figure)
fig = self.explainer.plot_shap_interaction("Sex_female", "Age")
self.assertIsInstance(fig, go.Figure)
fig = self.explainer.plot_shap_interaction("Gender", "Age")
self.assertIsInstance(fig, go.Figure)
fig = self.explainer.plot_shap_interaction("Age",
"Sex_female",
highlight_index=0)
self.assertIsInstance(fig, go.Figure)
def test_plot_pdp(self):
fig = self.explainer.plot_pdp("Age")
self.assertIsInstance(fig, go.Figure)
fig = self.explainer.plot_pdp("Gender")
self.assertIsInstance(fig, go.Figure)
fig = self.explainer.plot_pdp("Gender", index=0)
self.assertIsInstance(fig, go.Figure)
fig = self.explainer.plot_pdp("Age", index=0)
self.assertIsInstance(fig, go.Figure)
fig = self.explainer.plot_pdp("Age", X_row=self.explainer.X.iloc[[0]])
self.assertIsInstance(fig, go.Figure)
fig = self.explainer.plot_pdp("Age",
X_row=self.explainer.X_cats.iloc[[0]])
self.assertIsInstance(fig, go.Figure)
def test_yaml(self):
yaml = self.explainer.to_yaml()
self.assertIsInstance(yaml, str)
class LinearRegressionTests(unittest.TestCase):
def setUp(self):
X_train, y_train, X_test, y_test = titanic_fare()
self.test_len = len(X_test)
train_names, test_names = titanic_names()
_, self.names = titanic_names()
model = LinearRegression()
model.fit(X_train, y_train)
self.explainer = RegressionExplainer(model,
X_test,
y_test,
r2_score,
shap='linear',
cats=['Sex', 'Deck', 'Embarked'],
idxs=test_names,
units="$")
def test_explainer_len(self):
self.assertEqual(len(self.explainer), self.test_len)
def test_int_idx(self):
self.assertEqual(self.explainer.get_int_idx(self.names[0]), 0)
def test_random_index(self):
self.assertIsInstance(self.explainer.random_index(), int)
self.assertIsInstance(self.explainer.random_index(return_str=True),
str)
def test_preds(self):
self.assertIsInstance(self.explainer.preds, np.ndarray)
def test_pred_percentiles(self):
self.assertIsInstance(self.explainer.pred_percentiles, np.ndarray)
def test_permutation_importances(self):
self.assertIsInstance(self.explainer.permutation_importances,
pd.DataFrame)
self.assertIsInstance(self.explainer.permutation_importances_cats,
pd.DataFrame)
def test_metrics(self):
self.assertIsInstance(self.explainer.metrics(), dict)
self.assertIsInstance(self.explainer.metrics_markdown(), str)
def test_mean_abs_shap_df(self):
self.assertIsInstance(self.explainer.mean_abs_shap_df(), pd.DataFrame)
def test_top_interactions(self):
self.assertIsInstance(self.explainer.shap_top_interactions("Age"),
list)
self.assertIsInstance(
self.explainer.shap_top_interactions("Age", topx=4), list)
self.assertIsInstance(
self.explainer.shap_top_interactions("Age", cats=True), list)
self.assertIsInstance(
self.explainer.shap_top_interactions("Sex", cats=True), list)
def test_contrib_df(self):
self.assertIsInstance(self.explainer.contrib_df(0), pd.DataFrame)
self.assertIsInstance(self.explainer.contrib_df(0, cats=False),
pd.DataFrame)
self.assertIsInstance(self.explainer.contrib_df(0, topx=3),
pd.DataFrame)
def test_shap_base_value(self):
self.assertIsInstance(self.explainer.shap_base_value,
(np.floating, float))
def test_shap_values_shape(self):
self.assertTrue(
self.explainer.shap_values.shape == (len(self.explainer),
len(self.explainer.columns)))
def test_shap_values(self):
self.assertIsInstance(self.explainer.shap_values, np.ndarray)
self.assertIsInstance(self.explainer.shap_values_cats, np.ndarray)
def test_mean_abs_shap(self):
self.assertIsInstance(self.explainer.mean_abs_shap, pd.DataFrame)
self.assertIsInstance(self.explainer.mean_abs_shap_cats, pd.DataFrame)
def test_calculate_properties(self):
self.explainer.calculate_properties(include_interactions=False)
def test_pdp_result(self):
self.assertIsInstance(self.explainer.get_pdp_result("Age"),
pdpbox.pdp.PDPIsolate)
self.assertIsInstance(self.explainer.get_pdp_result("Sex"),
pdpbox.pdp.PDPIsolate)
self.assertIsInstance(self.explainer.get_pdp_result("Age", index=0),
pdpbox.pdp.PDPIsolate)
self.assertIsInstance(self.explainer.get_pdp_result("Sex", index=0),
pdpbox.pdp.PDPIsolate)
def test_get_dfs(self):
cols_df, shap_df, contribs_df = self.explainer.get_dfs()
self.assertIsInstance(cols_df, pd.DataFrame)
self.assertIsInstance(shap_df, pd.DataFrame)
self.assertIsInstance(contribs_df, pd.DataFrame)
def setUp(self):
model, X, y = get_skorch_regressor()
self.explainer = RegressionExplainer(model, X, y)
class LinearRegressionTests(unittest.TestCase):
def setUp(self):
X_train, y_train, X_test, y_test = titanic_fare()
self.test_len = len(X_test)
train_names, test_names = titanic_names()
_, self.names = titanic_names()
model = LinearRegression()
model.fit(X_train, y_train)
self.explainer = RegressionExplainer(
model,
X_test,
y_test,
shap='linear',
cats=[{
'Gender': ['Sex_female', 'Sex_male', 'Sex_nan']
}, 'Deck', 'Embarked'],
idxs=test_names,
units="$")
def test_explainer_len(self):
self.assertEqual(len(self.explainer), self.test_len)
def test_int_idx(self):
self.assertEqual(self.explainer.get_idx(self.names[0]), 0)
def test_random_index(self):
self.assertIsInstance(self.explainer.random_index(), int)
self.assertIsInstance(self.explainer.random_index(return_str=True),
str)
def test_preds(self):
self.assertIsInstance(self.explainer.preds, np.ndarray)
def test_pred_percentiles(self):
self.assertIsInstance(self.explainer.pred_percentiles(), np.ndarray)
def test_permutation_importances(self):
self.assertIsInstance(self.explainer.get_permutation_importances_df(),
pd.DataFrame)
def test_metrics(self):
self.assertIsInstance(self.explainer.metrics(), dict)
self.assertIsInstance(self.explainer.metrics_descriptions(), dict)
def test_mean_abs_shap_df(self):
self.assertIsInstance(self.explainer.get_mean_abs_shap_df(),
pd.DataFrame)
def test_top_interactions(self):
self.assertIsInstance(self.explainer.top_shap_interactions("Age"),
list)
self.assertIsInstance(
self.explainer.top_shap_interactions("Age", topx=4), list)
def test_contrib_df(self):
self.assertIsInstance(self.explainer.get_contrib_df(0), pd.DataFrame)
self.assertIsInstance(self.explainer.get_contrib_df(0, topx=3),
pd.DataFrame)
def test_shap_base_value(self):
self.assertIsInstance(self.explainer.shap_base_value(),
(np.floating, float))
def test_shap_values_shape(self):
self.assertTrue(self.explainer.get_shap_values_df().shape == (
len(self.explainer), len(self.explainer.merged_cols)))
def test_shap_values(self):
self.assertIsInstance(self.explainer.get_shap_values_df(),
pd.DataFrame)
def test_mean_abs_shap(self):
self.assertIsInstance(self.explainer.get_mean_abs_shap_df(),
pd.DataFrame)
def test_calculate_properties(self):
self.explainer.calculate_properties(include_interactions=False)
def test_pdp_df(self):
self.assertIsInstance(self.explainer.pdp_df("Age"), pd.DataFrame)
self.assertIsInstance(self.explainer.pdp_df("Gender"), pd.DataFrame)
self.assertIsInstance(self.explainer.pdp_df("Deck"), pd.DataFrame)
self.assertIsInstance(self.explainer.pdp_df("Age", index=0),
pd.DataFrame)
self.assertIsInstance(self.explainer.pdp_df("Gender", index=0),
pd.DataFrame)
class XGBRegressionTests(unittest.TestCase):
def setUp(self):
X_train, y_train, X_test, y_test = titanic_fare()
self.test_len = len(X_test)
train_names, test_names = titanic_names()
_, self.names = titanic_names()
model = XGBRegressor()
model.fit(X_train, y_train)
self.explainer = RegressionExplainer(model, X_test, y_test,
cats=[{'Gender': ['Sex_female', 'Sex_male', 'Sex_nan']},
'Deck', 'Embarked'],
units="$")
def test_preds(self):
self.assertIsInstance(self.explainer.preds, np.ndarray)
def test_permutation_importances(self):
self.assertIsInstance(self.explainer.get_permutation_importances_df(), pd.DataFrame)
def test_shap_base_value(self):
self.assertIsInstance(self.explainer.shap_base_value(), (np.floating, float))
def test_shap_values_shape(self):
self.assertTrue(self.explainer.get_shap_values_df().shape == (len(self.explainer), len(self.explainer.merged_cols)))
def test_shap_values(self):
self.assertIsInstance(self.explainer.get_shap_values_df(), pd.DataFrame)
def test_mean_abs_shap(self):
self.assertIsInstance(self.explainer.get_mean_abs_shap_df(), pd.DataFrame)
def test_calculate_properties(self):
self.explainer.calculate_properties(include_interactions=False)
def test_pdp_df(self):
self.assertIsInstance(self.explainer.pdp_df("Age"), pd.DataFrame)
self.assertIsInstance(self.explainer.pdp_df("Gender"), pd.DataFrame)
self.assertIsInstance(self.explainer.pdp_df("Deck"), pd.DataFrame)
self.assertIsInstance(self.explainer.pdp_df("Age", index=0), pd.DataFrame)
self.assertIsInstance(self.explainer.pdp_df("Gender", index=0), pd.DataFrame)
class ExternalSourceRegressionTests(unittest.TestCase):
def setUp(self):
X_train, y_train, X_test, y_test = titanic_fare()
model = RandomForestRegressor(n_estimators=50,
max_depth=4).fit(X_train, y_train)
X_test.reset_index(drop=True, inplace=True)
X_test.index = X_test.index.astype(str)
X_test1, y_test1 = X_test.iloc[:100], y_test.iloc[:100]
X_test2, y_test2 = X_test.iloc[100:], y_test.iloc[100:]
self.explainer = RegressionExplainer(model,
X_test1,
y_test1,
cats=['Sex', 'Deck'])
def index_exists_func(index):
return index in X_test2.index
def index_list_func():
# only returns first 50 indexes
return list(X_test2.index[:50])
def y_func(index):
idx = X_test2.index.get_loc(index)
return y_test2.iloc[[idx]]
def X_func(index):
idx = X_test2.index.get_loc(index)
return X_test2.iloc[[idx]]
self.explainer.set_index_exists_func(index_exists_func)
self.explainer.set_index_list_func(index_list_func)
self.explainer.set_X_row_func(X_func)
self.explainer.set_y_func(y_func)
def test_get_X_row(self):
self.assertIsInstance(self.explainer.get_X_row(0), pd.DataFrame)
self.assertIsInstance(self.explainer.get_X_row("0"), pd.DataFrame)
self.assertIsInstance(self.explainer.get_X_row("120"), pd.DataFrame)
self.assertIsInstance(self.explainer.get_X_row("150"), pd.DataFrame)
def test_get_shap_row(self):
self.assertIsInstance(self.explainer.get_shap_row(0), pd.DataFrame)
self.assertIsInstance(self.explainer.get_shap_row("0"), pd.DataFrame)
self.assertIsInstance(self.explainer.get_shap_row("120"), pd.DataFrame)
self.assertIsInstance(self.explainer.get_shap_row("150"), pd.DataFrame)
def test_get_y(self):
self.assertIsInstance(self.explainer.get_y(0), float)
self.assertIsInstance(self.explainer.get_y("0"), float)
self.assertIsInstance(self.explainer.get_y("120"), float)
self.assertIsInstance(self.explainer.get_y("150"), float)
def test_index_list(self):
index_list = self.explainer.get_index_list()
self.assertIn('100', index_list)
self.assertNotIn('160', index_list)
def test_index_exists(self):
self.assertTrue(self.explainer.index_exists("0"))
self.assertTrue(self.explainer.index_exists("100"))
self.assertTrue(self.explainer.index_exists("160"))
self.assertTrue(self.explainer.index_exists(0))
self.assertFalse(self.explainer.index_exists(-1))
self.assertFalse(self.explainer.index_exists(120))
self.assertFalse(self.explainer.index_exists("wrong index"))
class RegressionBunchTests(unittest.TestCase):
def setUp(self):
X_train, y_train, X_test, y_test = titanic_fare()
self.test_len = len(X_test)
train_names, test_names = titanic_names()
_, self.names = titanic_names()
model = RandomForestRegressor(n_estimators=5, max_depth=2)
model.fit(X_train, y_train)
self.explainer = RegressionExplainer(
model,
X_test,
y_test,
r2_score,
cats=[{
'Gender': ['Sex_female', 'Sex_male', 'Sex_nan']
}, 'Deck', 'Embarked'],
idxs=test_names)
def test_residuals(self):
self.assertIsInstance(self.explainer.residuals, pd.Series)
def test_prediction_result_markdown(self):
result_index = self.explainer.prediction_result_markdown(0)
self.assertIsInstance(result_index, str)
result_name = self.explainer.prediction_result_markdown(self.names[0])
self.assertIsInstance(result_name, str)
def test_metrics(self):
metrics_dict = self.explainer.metrics()
self.assertIsInstance(metrics_dict, dict)
self.assertTrue('root_mean_squared_error' in metrics_dict)
self.assertTrue('mean_absolute_error' in metrics_dict)
self.assertTrue('R-squared' in metrics_dict)
self.assertIsInstance(self.explainer.metrics_descriptions(), dict)
def test_plot_predicted_vs_actual(self):
fig = self.explainer.plot_predicted_vs_actual(logs=False)
self.assertIsInstance(fig, go.Figure)
fig = self.explainer.plot_predicted_vs_actual(logs=True)
self.assertIsInstance(fig, go.Figure)
fig = self.explainer.plot_predicted_vs_actual(log_x=True, log_y=True)
self.assertIsInstance(fig, go.Figure)
def test_plot_residuals(self):
fig = self.explainer.plot_residuals()
self.assertIsInstance(fig, go.Figure)
fig = self.explainer.plot_residuals(vs_actual=True)
self.assertIsInstance(fig, go.Figure)
fig = self.explainer.plot_residuals(residuals='ratio')
self.assertIsInstance(fig, go.Figure)
fig = self.explainer.plot_residuals(residuals='log-ratio')
self.assertIsInstance(fig, go.Figure)
fig = self.explainer.plot_residuals(residuals='log-ratio',
vs_actual=True)
self.assertIsInstance(fig, go.Figure)
def test_plot_residuals_vs_feature(self):
fig = self.explainer.plot_residuals_vs_feature("Age")
self.assertIsInstance(fig, go.Figure)
fig = self.explainer.plot_residuals_vs_feature("Age",
residuals='log-ratio')
self.assertIsInstance(fig, go.Figure)
fig = self.explainer.plot_residuals_vs_feature("Age", dropna=True)
self.assertIsInstance(fig, go.Figure)
fig = self.explainer.plot_residuals_vs_feature("Gender", points=False)
self.assertIsInstance(fig, go.Figure)
fig = self.explainer.plot_residuals_vs_feature("Gender", winsor=10)
self.assertIsInstance(fig, go.Figure)
def test_plot_y_vs_feature(self):
fig = self.explainer.plot_y_vs_feature("Age")
self.assertIsInstance(fig, go.Figure)
fig = self.explainer.plot_y_vs_feature("Age", dropna=True)
self.assertIsInstance(fig, go.Figure)
fig = self.explainer.plot_y_vs_feature("Gender", points=False)
self.assertIsInstance(fig, go.Figure)
fig = self.explainer.plot_y_vs_feature("Gender", winsor=10)
self.assertIsInstance(fig, go.Figure)
def test_plot_preds_vs_feature(self):
fig = self.explainer.plot_preds_vs_feature("Age")
self.assertIsInstance(fig, go.Figure)
fig = self.explainer.plot_preds_vs_feature("Age", dropna=True)
self.assertIsInstance(fig, go.Figure)
fig = self.explainer.plot_preds_vs_feature("Gender", points=False)
self.assertIsInstance(fig, go.Figure)
fig = self.explainer.plot_preds_vs_feature("Gender", winsor=10)
self.assertIsInstance(fig, go.Figure)
class RegressionBunchTests(unittest.TestCase):
def setUp(self):
X_train, y_train, X_test, y_test = titanic_fare()
self.test_len = len(X_test)
train_names, test_names = titanic_names()
_, self.names = titanic_names()
model = RandomForestRegressor(n_estimators=5, max_depth=2)
model.fit(X_train, y_train)
self.explainer = RegressionExplainer(model,
X_test,
y_test,
r2_score,
shap='tree',
cats=['Sex', 'Cabin', 'Embarked'],
idxs=test_names)
def test_residuals(self):
self.assertIsInstance(self.explainer.residuals, pd.Series)
def test_prediction_result_markdown(self):
result_index = self.explainer.prediction_result_markdown(0)
self.assertIsInstance(result_index, str)
result_name = self.explainer.prediction_result_markdown(self.names[0])
self.assertIsInstance(result_name, str)
def test_metrics(self):
metrics_dict = self.explainer.metrics()
self.assertIsInstance(metrics_dict, dict)
self.assertTrue('rmse' in metrics_dict)
self.assertTrue('mae' in metrics_dict)
self.assertTrue('R2' in metrics_dict)
def test_plot_predicted_vs_actual(self):
fig = self.explainer.plot_predicted_vs_actual(logs=False)
self.assertIsInstance(fig, go.Figure)
fig = self.explainer.plot_predicted_vs_actual(logs=True)
self.assertIsInstance(fig, go.Figure)
def test_plot_residuals(self):
fig = self.explainer.plot_residuals()
self.assertIsInstance(fig, go.Figure)
fig = self.explainer.plot_residuals(vs_actual=True)
self.assertIsInstance(fig, go.Figure)
fig = self.explainer.plot_residuals(ratio=True)
self.assertIsInstance(fig, go.Figure)
fig = self.explainer.plot_residuals(ratio=True, vs_actual=True)
self.assertIsInstance(fig, go.Figure)
def test_plot_residuals_vs_feature(self):
fig = self.explainer.plot_residuals_vs_feature("Age")
self.assertIsInstance(fig, go.Figure)
fig = self.explainer.plot_residuals_vs_feature("Age", ratio=True)
self.assertIsInstance(fig, go.Figure)
fig = self.explainer.plot_residuals_vs_feature("Age", dropna=True)
self.assertIsInstance(fig, go.Figure)
