def main():
records = readdata()
K = [5, 10, 50, 100]
KAnony = KAnonymity(records)
for k in K:
print('############# k-anonymity for k={} #############: \n'.format(k))
KAnony.anonymize(k=k)
print('\n')
def evaluate_laplace_mechanism(eps=[0.5, 1]):
"""
Evaluate for Laplace Mechanism
"""
recordsv0 = readdata()
recordsv1, recordsv2, recordsv3 = generate_data_for_laplace_mechanism(
recordsv0)
res1000 = {e: [] for e in eps}
# res4000 = {e: [] for e in eps}
rmse = {e: 0 for e in eps}
"""
evaluate for epsilon = 0.5 and 1 for 1000 queries
"""
printsent = [
'original data', 'data removed a record with the oldest age',
'data removed any record with age 26',
'data removed any record with the youghest age'
]
i = 0
for records in (recordsv0, recordsv1, recordsv2, recordsv3):
print('############ Processing for {} ############'.format(
printsent[i]))
i += 1
LampMec = laplace_mechanism.LaplaceMechanism(records)
for e in eps:
print('query 1000 results with epsilon = {}'.format(e))
res1000[e].append(LampMec.query_with_dp(e, querynum=1000))
# res4000[e].append(LampMec.query_with_dp(e, querynum=4000))
rmse[e] = LampMec.calc_distortion(
LampMec.query_with_dp(e, querynum=4000))
print('\n')
for e in eps:
print('############ Prove the {}-indistinguishable'.format(e))
for i in range(1, 4):
tmpresij, tmpresji = laplace_mechanism.prove_indistinguishable(
res1000[e][0], res1000[e][i])
print('** {} ** OVER ** {} **:'.format(printsent[0], printsent[i]))
print(tmpresij)
print('** {} ** OVER ** {} **:'.format(printsent[i], printsent[0]))
print(tmpresji)
print('exp^e = {}'.format(math.exp(e)))
print('\n')
print('############ Measure the distortion (RMSE) ############')
for e in eps:
print('RMSE for e = {}: {}'.format(e, rmse[e]))
print('Distortion of e=1 is smaller than e=0.5 ?: ',
True if rmse[1] <= rmse[0.5] else False)
del recordsv0
del recordsv1
del recordsv2
del recordsv3
def evaluate_exponential_mechanism(eps=[0.5, 1]):
"""
Evaulate for Exponential Mechanism
"""
recordsv0 = readdata()
recordsv1, recordsv2, recordsv3 = generate_data_for_exponential_mechanism(
recordsv0)
res1000 = {e: [] for e in eps}
# res4000 = {e: [] for e in eps}
dist = {e: 0 for e in eps}
"""
evaluate for epsilon = 0.5 and 1 for 1000 queries
"""
printsent = [
'original data', 'data removed a record with most frequent education',
'data removed a record with second most frequent education',
'data removed any record with the least frequent education'
]
i = 0
for records in (recordsv0, recordsv1, recordsv2, recordsv3):
print('############ Processing for {} ############'.format(
printsent[i]))
i += 1
ExpMe = exponential_mechanism.ExponentialMechanism(records)
for e in eps:
print('query 1000 results with epsilon = {}'.format(e))
res1000[e].append(ExpMe.query_with_dp(e, querynum=1000))
# res4000[e].append(LampMec.query_with_dp(e, querynum=4000))
dist[e] = ExpMe.calc_distortion(
ExpMe.query_with_dp(e, querynum=4000))
print('\n')
for e in eps:
print('############ Prove the {}-indistinguishable'.format(e))
for i in range(1, 4):
tmpresij, tmpresji = exponential_mechanism.prove_indistinguishable(
res1000[e][0], res1000[e][i])
print('** {} ** OVER ** {} **:'.format(printsent[0], printsent[i]))
print(tmpresij)
print('** {} ** OVER ** {} **:'.format(printsent[i], printsent[0]))
print(tmpresji)
print('exp^e = {}'.format(math.exp(e)))
print('\n')
print('############ Measure the distortion (1-precision) ############')
for e in eps:
print('distortion for e = {}: {}'.format(e, dist[e]))
print('Distortion of e=1 is smaller than e=0.5 ?: ',
True if dist[1] <= dist[0.5] else False)
for val1, val2 in zip(queryres1, queryres2):
count1[math.floor((val1 - minval + 1) /
((maxval - minval + 1) / bucketnum)) - 1] += 1
count2[math.floor((val2 - minval + 1) //
((maxval - minval + 1) / bucketnum)) - 1] += 1
prob1 = list(map(lambda x: x / len(queryres1), count1))
prob2 = list(map(lambda x: x / len(queryres2), count2))
res1overres2 = sum(p1 / p2
for p1, p2 in zip(prob1, prob2) if p2 != 0) / bucketnum
res2overres1 = sum(p2 / p1
for p1, p2 in zip(prob1, prob2) if p1 != 0) / bucketnum
return res1overres2, res2overres1
if __name__ == "__main__":
records = readdata()
v1, v2, v3 = generate_data_for_laplace_mechanism(records)
LapMe = LaplaceMechanism(records)
res1 = LapMe.query_with_dp(0.5, 1000)
print(res1)
print(LapMe.calc_groundtruth())
print(LapMe.calc_distortion(LapMe.query_with_dp(1, 1000)))
LapMe2 = LaplaceMechanism(v1)
res2 = LapMe2.query_with_dp(0.5, 1000)
print(LapMe.calc_distortion(res1))
print(LapMe2.calc_distortion(res2))
print(prove_indistinguishable(res1, res2))
print(prove_indistinguishable(res2, res1))
print(math.exp(0.5))