def test_output_format_checked():
input_features, y_true = pd.DataFrame(data=[range(15)]), pd.Series(
range(15))
with pytest.raises(
ValueError,
match=
"Parameter output_format must be either text or dict. Received bar"
):
explain_predictions(None, input_features, output_format="bar")
with pytest.raises(
ValueError,
match=
"Parameter output_format must be either text or dict. Received xml"
):
explain_prediction(None,
input_features=input_features,
training_data=None,
output_format="xml")
input_features, y_true = pd.DataFrame(data=range(15)), pd.Series(range(15))
with pytest.raises(
ValueError,
match=
"Parameter output_format must be either text or dict. Received foo"
):
explain_predictions_best_worst(None,
input_features,
y_true=y_true,
output_format="foo")
def test_output_format_checked():
input_features, y_true = pd.DataFrame(data=[range(15)]), pd.Series(
range(15))
with pytest.raises(
ValueError,
match=
"Parameter output_format must be either text, dict, or dataframe. Received bar"
):
explain_predictions(pipeline=MagicMock(),
input_features=input_features,
y=None,
indices_to_explain=0,
output_format="bar")
with pytest.raises(
ValueError,
match=
"Parameter output_format must be either text, dict, or dataframe. Received xml"
):
explain_prediction(pipeline=MagicMock(),
input_features=input_features,
y=None,
index_to_explain=0,
output_format="xml")
input_features, y_true = pd.DataFrame(data=range(15)), pd.Series(range(15))
with pytest.raises(
ValueError,
match=
"Parameter output_format must be either text, dict, or dataframe. Received foo"
):
explain_predictions_best_worst(pipeline=MagicMock(),
input_features=input_features,
y_true=y_true,
output_format="foo")
def test_explain_prediction_value_error(test_features):
with pytest.raises(
ValueError,
match=
"features must be stored in a dataframe or datatable with exactly one row."
):
explain_prediction(None,
input_features=test_features,
training_data=None)
def test_explain_prediction_errors():
with pytest.raises(ValueError,
match="Explained indices should be between"):
explain_prediction(MagicMock(),
pd.DataFrame({"a": [0, 1, 2, 3, 4]}),
y=None,
index_to_explain=5)
with pytest.raises(ValueError,
match="Explained indices should be between"):
explain_prediction(MagicMock(),
pd.DataFrame({"a": [0, 1, 2, 3, 4]}),
y=None,
index_to_explain=-1)
def test_explain_prediction(mock_normalize_shap_values,
mock_compute_shap_values, problem_type,
output_format, shap_values, normalized_shap_values,
answer, input_type):
mock_compute_shap_values.return_value = shap_values
mock_normalize_shap_values.return_value = normalized_shap_values
pipeline = MagicMock()
pipeline.problem_type = problem_type
pipeline.classes_ = ["class_0", "class_1", "class_2"]
# By the time we call transform, we are looking at only one row of the input data.
pipeline.compute_estimator_features.return_value = pd.DataFrame({
"a": [10],
"b": [20],
"c": [30],
"d": [40]
})
features = pd.DataFrame({"a": [1], "b": [2]})
training_data = pd.DataFrame()
if input_type == "ww":
features = ww.DataTable(features)
training_data = ww.DataTable(training_data)
table = explain_prediction(pipeline,
features,
output_format=output_format,
top_k=2,
training_data=training_data)
if isinstance(table, str):
compare_two_tables(table.splitlines(), answer)
else:
assert table == answer
def test_explain_prediction(mock_normalize_shap_values,
mock_compute_shap_values, problem_type,
output_format, shap_values, normalized_shap_values,
answer, input_type):
mock_compute_shap_values.return_value = shap_values
mock_normalize_shap_values.return_value = normalized_shap_values
pipeline = MagicMock()
pipeline.problem_type = problem_type
pipeline.classes_ = ["class_0", "class_1", "class_2"]
# By the time we call transform, we are looking at only one row of the input data.
pipeline.compute_estimator_features.return_value = ww.DataTable(
pd.DataFrame({
"a": [10],
"b": [20],
"c": [30],
"d": [40]
}))
features = pd.DataFrame({"a": [1], "b": [2]})
if input_type == "ww":
features = ww.DataTable(features)
with warnings.catch_warnings(record=True) as warn:
warnings.simplefilter("always")
table = explain_prediction(pipeline,
features,
y=None,
output_format=output_format,
index_to_explain=0,
top_k_features=2)
assert str(warn[0].message).startswith(
"The explain_prediction function will be deleted in the next release"
)
if isinstance(table, str):
compare_two_tables(table.splitlines(), answer)
elif isinstance(table, pd.DataFrame):
pd.testing.assert_frame_equal(table, answer)
else:
assert table == answer