thesis-medical-data-collection/kanonymity/datasets/categorical.py

# [is_quasi_identifier, is_categorical]
ADULT_ATTRIBUTES = {'sex': [True, True],
                    'age': [True, False],
                    'race': [True, True],
                    'marital-status': [True, True],
                    'education': [True, True],
                    'native-country': [True, True],
                    'workclass': [True, True],
                    'occupation': [True, True],
                    'salary-class': [False],
                    'ID': [False]
}

DATASET_ATTRIBUTES_DICT = {
    'adult': ADULT_ATTRIBUTES,
}

QI_INDEX_DICT = {
    'adult': [1, 2, 3, 4, 5, 6, 7, 8],
}