def test_re_identification_risk_to_dataframe_shape(self):
risk_profile = RiskProfile(self.risk_profile_response)
df = risk_profile.re_identification_risk_dataframe()
self.assertEqual(
self.risk_profile_response["reIdentificationRisk"]["measures"]
["records_affected_by_highest_prosecutor_risk"],
df["records_affected_by_highest_prosecutor_risk"][0])
def test_equality(self):
risk_profile_1 = RiskProfile(self.risk_profile_response)
risk_profile_2 = RiskProfile(self.risk_profile_response)
self.assertEqual(risk_profile_1, risk_profile_2)
risk_profile_2._re_identification_of_risk[
"estimated_prosecutor_risk"] = 50.0
self.assertNotEqual(risk_profile_1, risk_profile_2)
def test_attacker_success_rate_property(self):
expected = {
'Prosecutor_attacker_success_rate': 1.0,
'Marketer_attacker_success_rate': 1.0,
'Journalist_attacker_success_rate': 1.0
}
risk_profile = RiskProfile(self.risk_profile_response)
self.assertEqual(expected, risk_profile.attacker_success_rate)
def risk_profile(self, dataset: Dataset) -> RiskProfile:
"""
Creates a risk profile for a provided Dataset
RiskProfile contains:
- re-identifiaction risks
- distributed risk
:param dataset: Dataset to create a risk profile for
:return: RiskProfile
"""
analyze_request = self._risk_profile_payload(dataset)
response = self._risk_profile(analyze_request)
metric_dict = json.loads(response.text)
return RiskProfile(metric_dict)
def _anonymize_result(self, response):
"""
Creates the result to be delivered back to the caller
:param response:
:return:
"""
json_string = response.text
response_dict = json.loads(json_string)
attributes = self._attributes(response_dict)
dataset = Dataset(response_dict["anonymizeResult"]["data"], attributes)
risk_profile = RiskProfile(response_dict["riskProfile"])
anon_status = response_dict["anonymizeResult"]["anonymizationStatus"]
anonymization_metrics = response_dict["anonymizeResult"]["metrics"]
return AnonymizeResult._from_response(dataset, risk_profile,
anonymization_metrics,
anon_status)
def setUp(self):
self.test_data = [['id', 'name'], ['0', 'Viktor'], ['1', 'Jerry']]
self.test_attribute_type_mapping = {
'id': AttributeType.IDENTIFYING,
'name': AttributeType.QUASIIDENTIFYING
}
self.test_dataset = Dataset(self.test_data,
self.test_attribute_type_mapping)
self.risk_profile_response = {
"reIdentificationRisk": {
"measures": {
"measure_value":
"[%]",
"Prosecutor_attacker_success_rate":
"98.72",
"records_affected_by_highest_prosecutor_risk":
"97.46000000000001",
"sample_uniques":
"97.46000000000001",
"estimated_prosecutor_risk":
"100.0",
"population_model":
"PITMAN",
"highest_journalist_risk":
"100.0",
"records_affected_by_lowest_risk":
"0.06",
"estimated_marketer_risk":
"98.72000000000001",
"Journalist_attacker_success_rate":
"98.72",
"highest_prosecutor_risk":
"100.0",
"estimated_journalist_risk":
"100.0",
"lowest_risk":
"33.33333333333333",
"Marketer_attacker_success_rate":
"98.72",
"average_prosecutor_risk":
"98.72000000000001",
"records_affected_by_highest_journalist_risk":
"97.46000000000001",
"population_uniques":
"39.64593493418713",
"quasi_identifiers": [
"Innvandrerbakgrunn", "Ytelse", "Innsatsgruppe",
"Ledighetsstatus"
]
}
},
"distributionOfRisk": {
"riskIntervalList": [{
"interval": "]50,100]",
"recordsWithRiskWithinInteval": 0.9746,
"recordsWithMaxmalRiskWithinInterval": 1.0
}, {
"interval":
"]33.4,50]",
"recordsWithRiskWithinInteval":
0.0248,
"recordsWithMaxmalRiskWithinInterval":
0.0254
}, {
"interval":
"]25,33.4]",
"recordsWithRiskWithinInteval":
0.0006,
"recordsWithMaxmalRiskWithinInterval":
0.0006
}, {
"interval": "]20,25]",
"recordsWithRiskWithinInteval": 0.0,
"recordsWithMaxmalRiskWithinInterval": 0.0
}, {
"interval": "]16.7,20]",
"recordsWithRiskWithinInteval": 0.0,
"recordsWithMaxmalRiskWithinInterval": 0.0
}, {
"interval": "]14.3,16.7]",
"recordsWithRiskWithinInteval": 0.0,
"recordsWithMaxmalRiskWithinInterval": 0.0
}, {
"interval": "]12.5,14.3]",
"recordsWithRiskWithinInteval": 0.0,
"recordsWithMaxmalRiskWithinInterval": 0.0
}, {
"interval": "]10,12.5]",
"recordsWithRiskWithinInteval": 0.0,
"recordsWithMaxmalRiskWithinInterval": 0.0
}, {
"interval": "]9,10]",
"recordsWithRiskWithinInteval": 0.0,
"recordsWithMaxmalRiskWithinInterval": 0.0
}, {
"interval": "]8,9]",
"recordsWithRiskWithinInteval": 0.0,
"recordsWithMaxmalRiskWithinInterval": 0.0
}, {
"interval": "]7,8]",
"recordsWithRiskWithinInteval": 0.0,
"recordsWithMaxmalRiskWithinInterval": 0.0
}, {
"interval": "]6,7]",
"recordsWithRiskWithinInteval": 0.0,
"recordsWithMaxmalRiskWithinInterval": 0.0
}, {
"interval": "]5,6]",
"recordsWithRiskWithinInteval": 0.0,
"recordsWithMaxmalRiskWithinInterval": 0.0
}, {
"interval": "]4,5]",
"recordsWithRiskWithinInteval": 0.0,
"recordsWithMaxmalRiskWithinInterval": 0.0
}, {
"interval": "]3,4]",
"recordsWithRiskWithinInteval": 0.0,
"recordsWithMaxmalRiskWithinInterval": 0.0
}, {
"interval": "]2,3]",
"recordsWithRiskWithinInteval": 0.0,
"recordsWithMaxmalRiskWithinInterval": 0.0
}, {
"interval": "]1,2]",
"recordsWithRiskWithinInteval": 0.0,
"recordsWithMaxmalRiskWithinInterval": 0.0
}, {
"interval": "]0.1,1]",
"recordsWithRiskWithinInteval": 0.0,
"recordsWithMaxmalRiskWithinInterval": 0.0
}, {
"interval": "]0.01,0.1]",
"recordsWithRiskWithinInteval": 0.0,
"recordsWithMaxmalRiskWithinInterval": 0.0
}, {
"interval": "]0.001,0.01]",
"recordsWithRiskWithinInteval": 0.0,
"recordsWithMaxmalRiskWithinInterval": 0.0
}, {
"interval": "]0.0001,0.001]",
"recordsWithRiskWithinInteval": 0.0,
"recordsWithMaxmalRiskWithinInterval": 0.0
}, {
"interval": "]1e-5,0.0001]",
"recordsWithRiskWithinInteval": 0.0,
"recordsWithMaxmalRiskWithinInterval": 0.0
}, {
"interval": "]1e-6,1e-5]",
"recordsWithRiskWithinInteval": 0.0,
"recordsWithMaxmalRiskWithinInterval": 0.0
}, {
"interval": "]0,1e-6]",
"recordsWithRiskWithinInteval": 0.0,
"recordsWithMaxmalRiskWithinInterval": 0.0
}]
}
}
self.test_riskprofile = RiskProfile(self.risk_profile_response)
self.test_anonymize_result = AnonymizeResult(self.test_data,
self.test_riskprofile,
self.test_metrics)
def risk_profile() -> RiskProfile:
raw_risk_profile = analyze_response()
return RiskProfile(raw_risk_profile)
def test_population_model_property(self):
expected = "ZAYATZ"
risk_profile = RiskProfile(self.risk_profile_response)
self.assertEqual(expected, risk_profile.population_model)
def test_quasi_indentifers_property(self):
expected = ['zipcode']
risk_profile = RiskProfile(self.risk_profile_response)
self.assertEqual(expected, risk_profile.quasi_identifiers)
def test_re_identification_risk_to_dataframe__column_types(self):
risk_profile = RiskProfile(self.risk_profile_response)
df = risk_profile.re_identification_risk_dataframe()
for d_type in df.dtypes.tolist():
self.assertEqual(d_type, dtype("float64"))
def test_distribution_of_risk_to_dataframe_shape(self):
risk_profile = RiskProfile(self.risk_profile_response)
df = risk_profile.distribution_of_risk_dataframe()
self.assertEqual(
self.risk_profile_response["distributionOfRisk"]
["riskIntervalList"][0]["interval"], df["interval"][0])
def test_to_dataframe(self):
risk_profile = RiskProfile(self.risk_profile_response)
df = risk_profile.re_identification_risk_dataframe()
self.assertIsInstance(df, DataFrame)
def test_hash(self):
risk_profile_1 = RiskProfile(self.risk_profile_response)
risk_profile_2 = RiskProfile(self.risk_profile_response)
test_set = {risk_profile_1, risk_profile_2}
self.assertEqual(1, len(test_set))
def test_init(self):
RiskProfile(self.risk_profile_response)