def test_transform():
data_name = "adult"
transformed_data = OnlineCatalog(
data_name, scaling_method="MinMax", encoding_method="OneHot_drop_binary"
)
raw_data = OnlineCatalog(
data_name, scaling_method="Identity", encoding_method="Identity"
)
# sort columns as order could be different
assert_frame_equal(
transformed_data.inverse_transform(transformed_data.df).sort_index(axis=1),
raw_data.df.sort_index(axis=1),
check_dtype=False,
)
assert_frame_equal(
transformed_data.transform(raw_data.df).sort_index(axis=1),
transformed_data.df.sort_index(axis=1),
check_dtype=False,
)
assert_frame_equal(
transformed_data.transform(
transformed_data.inverse_transform(transformed_data.df)
).sort_index(axis=1),
transformed_data.df.sort_index(axis=1),
check_dtype=False,
)
assert_frame_equal(
transformed_data.inverse_transform(
transformed_data.transform(raw_data.df)
).sort_index(axis=1),
raw_data.df.sort_index(axis=1),
check_dtype=False,
)
def test_constraint_violations():
# Build data and mlmodel
data_name = "adult"
data = OnlineCatalog(data_name)
# get factuals
columns = [
"age",
"workclass",
"fnlwgt",
"education-num",
"marital-status",
"occupation",
"relationship",
"race",
"sex",
"capital-gain",
"capital-loss",
"hours-per-week",
"native-country",
"income",
]
test_factual = [
[
39,
"Non-Private",
77516,
13,
"Non-Married",
"Managerial-Specialist",
"Non-Husband",
"White",
"Male",
2174,
0,
40,
"US",
0,
],
[
50,
"Non-Private",
83311,
13,
"Married",
"Managerial-Specialist",
"Husband",
"White",
"Male",
0,
0,
13,
"US",
0,
],
[
38,
"Private",
215646,
9,
"Non-Married",
"Other",
"Non-Husband",
"White",
"Male",
0,
0,
40,
"US",
0,
],
[
53,
"Private",
234721,
7,
"Married",
"Other",
"Husband",
"Non-White",
"Male",
0,
0,
40,
"US",
0,
],
[
28,
"Private",
338409,
13,
"Married",
"Managerial-Specialist",
"Non-Husband",
"Non-White",
"Female",
0,
0,
40,
"Non-US",
0,
],
]
test_factual = pd.DataFrame(
test_factual,
columns=columns,
)
test_counterfactual = [
[
45,
"Non-Private",
77516,
13,
"Non-Married",
"Managerial-Specialist",
"Non-Husband",
"White",
"Female",
2174,
0,
40,
"US",
0,
],
[
50,
"Non-Private",
83311,
13,
"Married",
"Managerial-Specialist",
"Husband",
"White",
"Male",
0,
0,
13,
"US",
0,
],
[
18,
"Private",
215646,
9,
"Non-Married",
"Other",
"Non-Husband",
"White",
"Male",
0,
0,
40,
"US",
0,
],
[
53,
"Private",
234721,
7,
"Married",
"Other",
"Husband",
"Non-White",
"Male",
0,
0,
40,
"US",
0,
],
[
28,
"Private",
338409,
13,
"Married",
"Managerial-Specialist",
"Non-Husband",
"Non-White",
"Male",
0,
0,
40,
"Non-US",
0,
],
]
test_counterfactual = pd.DataFrame(
test_counterfactual,
columns=columns,
)
test_counterfactual = data.transform(test_counterfactual)
test_factual = data.transform(test_factual)
expected = [[2], [0], [1], [0], [1]]
actual = constraint_violation(data, test_counterfactual, test_factual)
assert expected == actual