def benchIdentitySynthesizer():
data = np.loadtxt('down_data.csv', delimiter=',', skiprows=1)
with open('generated_metadata.json') as data_file:
data2 = json.load(data_file)
categorical_columns = list()
ordinal_columns = list()
for column_idx, column in enumerate(data2['columns']):
if column['type'] == CATEGORICAL:
print(column)
print('Classified as Categorical')
categorical_columns.append(column_idx)
elif column['type'] == ORDINAL:
ordinal_columns.append(column_idx)
print(column)
print('Classified as Ordinal')
synthesizer = IdentitySynthesizer()
synthesizer.fit(data, categorical_columns, ordinal_columns)
sampled = synthesizer.sample(300)
scores = benchmark(synthesizer.fit_sample)
scores = scores.append(synthesizer.fit_sample)
scores = scores.append(synthesizer.fit_sample)
print('\nEvaluation Scores from evaluate function:\n')
print(scores)
scores['Synth'] = 'IdentitySynthesizer'
scores.to_csv('IdentityBench.csv')
def identity_synthesis(json = 'adult'):
data, categorical_columns, ordinal_columns = load_dataset(json)
synthesizer = IdentitySynthesizer()
synthesizer.fit(data, categorical_columns, ordinal_columns)
sampled = synthesizer.sample(10)
print(sampled)
np.savetxt('test.txt', sampled, delimiter=',')
return sampled
def identity_benchmark(json = 'adult'):
train, test, meta, categoricals, ordinals = load_dataset(json, benchmark=True)
synthesizer = IdentitySynthesizer()
synthesizer.fit(train, categoricals, ordinals)
sampled = synthesizer.sample(300)
print('Sampled Data for 300 records\n')
scores = evaluate(train, test, sampled, meta)
print('\nEvaluation Scores from evaluate function:\n')
return scores
def identity():
data = np.loadtxt('down_data.csv', delimiter=',', skiprows=1)
categorical_columns = []
ordinal_columns = []
synthesizer = IdentitySynthesizer()
synthesizer.fit(data, categorical_columns, ordinal_columns)
sampled = synthesizer.sample(4000)
np.savetxt("41_identity.csv", sampled, delimiter=",")
data = pd.read_csv('41_identity.csv', header=None)
return data
def identitySynth():
import numpy as np
from sdgym.constants import CATEGORICAL, ORDINAL
import json
from sdgym.synthesizers import IdentitySynthesizer
from configparser import ConfigParser
import boto3
with open('generated_metadata.json') as data_file:
data = json.load(data_file)
categorical_columns = list()
ordinal_columns = list()
for column_idx, column in enumerate(data['columns']):
if column['type'] == CATEGORICAL:
print(column)
print('Classified as Categorical')
categorical_columns.append(column_idx)
elif column['type'] == ORDINAL:
ordinal_columns.append(column_idx)
print(column)
print('Classified as Ordinal')
# return categorical_columns, ordinal_columns
data = np.loadtxt('down_data.csv', delimiter=',', skiprows=1)
synthesizer = IdentitySynthesizer()
synthesizer.fit(data, categorical_columns, ordinal_columns)
sampled = synthesizer.sample(4000)
np.savetxt("43_identity.csv", sampled, delimiter=",")
print(sampled)
print('Data Synthesized using Identity synthesizer')
s3 = boto3.resource(
's3',
aws_access_key_id='',
aws_secret_access_key='',
)
s3.Bucket('csye7245-1').upload_file('43_identity.csv',
'synth/43_identity.csv')
print('Synthesized(Identity) Data Uploaded to S3')
def benchIdentitySynthesizer():
from sdgym.synthesizers import IdentitySynthesizer
from sdgym.evaluate import evaluate
from sdgym.data import load_dataset
train, test, meta, categoricals, ordinals = load_dataset('adult',
benchmark=True)
synthesizer = IdentitySynthesizer()
synthesizer.fit(train, categoricals, ordinals)
sampled = synthesizer.sample(300)
scores = evaluate(train, test, sampled, meta)
#scores = scores.append(evaluate(train, test, sampled, meta))
#scores = scores.append(evaluate(train, test, sampled, meta))
print('\nEvaluation Scores from evaluate function:\n')
print(scores)
scores['Synth'] = 'IdentitySynthesizer'
scores.to_csv('IdentityBench.csv')
def identity_synthesis(json='adult'):
data, categorical_columns, ordinal_columns = load_dataset(json)
synthesizer = IdentitySynthesizer()
synthesizer.fit(data, categorical_columns, ordinal_columns)
sampled = synthesizer.sample(10)
sampled
"census",
"child",
"covtype",
"credit",
"grid",
"gridr",
"insurance",
"intrusion",
"mnist12",
"mnist28",
"news",
"ring"
]
synthesizer = IdentitySynthesizer()
MANUAL_TESTS = os.environ.get('MANUAL_TESTS', 'false') == 'true'
MESSAGE = 'set environ variable MANUAL_TESTS="true" to execute'
def test_adult():
benchmark(synthesizer.fit_sample, repeat=1, datasets=['adult'])
@unittest.skipUnless(MANUAL_TESTS, MESSAGE)
def test_alarm():
benchmark(synthesizer.fit_sample, repeat=1, datasets=['alarm'])
def evaluatefun():
from sdgym.synthesizers import IndependentSynthesizer
from sdgym.evaluate import evaluate
data = np.loadtxt('down_data.csv', delimiter=',', skiprows=1)
with open('generated_metadata.json') as data_file:
data2 = json.load(data_file)
categorical_columns = list()
ordinal_columns = list()
for column_idx, column in enumerate(data2['columns']):
if column['type'] == CATEGORICAL:
print(column)
print('Classified as Categorical')
categorical_columns.append(column_idx)
elif column['type'] == ORDINAL:
ordinal_columns.append(column_idx)
print(column)
print('Classified as Ordinal')
synthesizer = IdentitySynthesizer()
synthesizer.fit(data, categorical_columns, ordinal_columns)
scores = benchmark(synthesizer.fit_sample)
scores['Synth'] = 'IdentitySynthesizer'
synthesizer = UniformSynthesizer()
synthesizer.fit(data, categorical_columns, ordinal_columns)
scores2 = benchmark(synthesizer.fit_sample)
scores2['Synth'] = 'Uniform'
synthesizer = IndependentSynthesizer()
synthesizer.fit(data, categorical_columns, ordinal_columns)
scores3 = benchmark(synthesizer.fit_sample)
scores3['Synth'] = 'Identity'
synthesizer = CLBNSynthesizer()
synthesizer.fit(data, categorical_columns, ordinal_columns)
scores4 = benchmark(synthesizer.fit_sample)
scores4['Synth'] = 'CLBN'
print('\nEvaluation Scores from evaluate function:\n')
result = scores.append(scores2)
result = result.append(scores3)
result = result.append(scores4)
a = (result[result['accuracy'] == result['accuracy'].max()])
st.write('Best Performing Synthsizer: ' + str(a['Synth'].item()))
st.write('Accuracy: ' + str(a['accuracy'].item()))
st.dataframe(result)