apunkt/ai-privacy-toolkit
1
0
Fork 0
mirror of https://github.com/IBM/ai-privacy-toolkit.git synced 2026-06-08 15:05:13 +02:00
Code Issues Projects Releases Packages Wiki Activity Actions

Second test (pandas)

Browse source 
Signed-off-by: abigailt <abigailt@il.ibm.com>
This commit is contained in:
abigailt 2023-11-15 08:27:17 -05:00
parent 2a657388af
commit e7ee42fdc8
1 changed files with 21 additions and 13 deletions
Download patch file Download diff file Expand all files Collapse all files

34
tests/test_minimizer.py
View file

@ -962,7 +962,7 @@ def test_minimizer_ndarray_one_hot():
def test_anonymize_pandas_one_hot(): def test_anonymize_pandas_one_hot():
feature_names = ["age", "gender_M", "gender_F", "height"] features = ["age", "gender_M", "gender_F", "height"]
x_train = np.array([[23, 0, 1, 165], x_train = np.array([[23, 0, 1, 165],
[45, 0, 1, 158], [45, 0, 1, 158],
[56, 1, 0, 123], [56, 1, 0, 123],
@ -975,25 +975,33 @@ def test_anonymize_pandas_one_hot():
[24, 1, 0, 181], [24, 1, 0, 181],
[18, 1, 0, 190]]) [18, 1, 0, 190]])
y_train = np.array([1, 1, 0, 1, 0, 0, 1, 1, 0, 1, 0]) y_train = np.array([1, 1, 0, 1, 0, 0, 1, 1, 0, 1, 0])
x_train = pd.DataFrame(x_train, columns=feature_names) x_train = pd.DataFrame(x_train, columns=features)
y_train = pd.Series(y_train) y_train = pd.Series(y_train)
model = DecisionTreeClassifier() model = DecisionTreeClassifier()
model.fit(x_train, y_train) model.fit(x_train, y_train)
pred = model.predict(x_train) predictions = model.predict(x_train)
k = 10
QI = ["age", "gender_M", "gender_F"] QI = ["age", "gender_M", "gender_F"]
QI_slices = [["gender_M", "gender_F"]] QI_slices = [["gender_M", "gender_F"]]
anonymizer = Anonymize(k, QI, train_only_QI=True, quasi_identifer_slices=QI_slices) target_accuracy = 0.7
anon = anonymizer.anonymize(ArrayDataset(x_train, pred)) gen = GeneralizeToRepresentative(model, target_accuracy=target_accuracy, feature_slices=QI_slices,
assert (anon.loc[:, QI].drop_duplicates().shape[0] < x_train.loc[:, QI].drop_duplicates().shape[0]) features_to_minimize=QI)
assert (anon.loc[:, QI].value_counts().min() >= k) gen.fit(dataset=ArrayDataset(x_train, predictions))
np.testing.assert_array_equal(anon.drop(QI, axis=1), x_train.drop(QI, axis=1)) transformed = gen.transform(dataset=ArrayDataset(x_train))
anonymized_slice = anon.loc[:, QI_slices[0]] gener = gen.generalizations
assert ((np.sum(anonymized_slice, axis=1) == 1).all()) expected_generalizations = {'categories': {}, 'category_representatives': {},
assert ((np.max(anonymized_slice, axis=1) == 1).all()) 'range_representatives': {'age': [34.5]},
assert ((np.min(anonymized_slice, axis=1) == 0).all()) 'ranges': {'age': [34.5]}, 'untouched': ['height', 'gender_M', 'gender_F']}
compare_generalizations(gener, expected_generalizations)
check_features(features, expected_generalizations, transformed, x_train, True)
ncp = gen.ncp.transform_score
check_ncp(ncp, expected_generalizations)
rel_accuracy = model.score(transformed, predictions)
assert ((rel_accuracy >= target_accuracy) or (target_accuracy - rel_accuracy) <= ACCURACY_DIFF)
def test_keras_model(): def test_keras_model():