data_pub, _ = randomKway(args.dataset_pub,
args.workload,
args.marginal,
seed=args.workload_seed,
proj=proj,
filter=filter_pub)
N_pub = int(args.pub_frac * data_pub.df.shape[0])
data_pub, A_init = get_pub_dataset(data_pub, args.pub_frac, args.frac_seed)
print('workload: ', len(workloads))
print('num queries: ', query_manager.num_queries)
print('A:', A_init.shape)
real_answers = query_manager.get_answer(data)
query_manager.setup_query_attr(save_dir=save_dir_query)
query_manager.setup_xy(data_pub, save_dir=save_dir_xy)
fake_answers = query_manager.get_answer_weights(A_init)
init_error = np.abs(real_answers - fake_answers).max()
result_cols = {
'marginal': args.marginal,
'num_workloads': len(workloads),
'workload_seed': args.workload_seed,
'num_queries': query_manager.num_queries,
'dataset_pub': args.dataset_pub,
'state_pub': args.state_pub,
'pub_frac': args.pub_frac,
'frac_seed': args.frac_seed,
'priv_size': N,
'pub_size': N_pub,
}
df_support = data_support.df
prng = np.random.RandomState(args.support_seed)
idxs = prng.choice(df_support.index.values, size=args.support_size, replace=False)
df_support = df_support.loc[idxs].reset_index(drop=True)
data_support = Dataset(df_support, data_support.domain)
A_init = np.ones(len(df_support))
A_init = A_init / len(A_init)
print('workload: ', len(workloads))
print('num queries: ', query_manager.num_queries)
print('A:', A_init.shape)
# get answers and initial error
real_answers = query_manager.get_answer(data, concat=False)
query_manager.setup_query_attr(save_dir=save_dir_query)
query_manager.setup_xy(data_support, save_dir=save_dir_xy)
fake_answers = query_manager.get_answer_weights(A_init, concat=False)
init_errors = util.get_errors(real_answers, fake_answers)
delta = 1.0 / N ** 2
rho = cdp_rho(args.epsilon, delta)
eps0 = (2 * rho) ** 0.5 / (2 * args.T) ** 0.5
result_cols = {'adult_seed': [args.adult_seed],
'marginal': [args.marginal],
'num_workloads': [len(workloads)],
'workload_seed': [args.workload_seed],
'num_queries': [query_manager.num_queries],
'priv_size': [N],
'support_size': [args.support_size],
'support_seed': [args.support_seed]
filter_private, filter_pub = get_filters(args)
data, workloads = randomKway(args.dataset,
args.workload,
args.marginal,
seed=args.workload_seed,
proj=proj,
filter=filter_private)
query_manager = QueryManager(data.domain, workloads)
N = data.df.shape[0]
data_support, A_init = get_support(data)
# get answers and initial error
real_answers = query_manager.get_answer(data)
query_manager.setup_xy(data_support)
fake_answers = query_manager.get_answer_weights(A_init)
init_error = np.abs(real_answers - fake_answers).max()
A_avg, A_last = generate_nondp(data_support,
real_answers,
A_init,
query_manager,
early_stopping=args.early_stopping,
return_last=True)
fake_answers = query_manager.get_answer_weights(A_avg)
max_error_avg = np.abs(real_answers - fake_answers).max()
fake_answers = query_manager.get_answer_weights(A_last)
max_error_last = np.abs(real_answers - fake_answers).max()
pd_states = pd.read_csv('Datasets/{}.csv'.format(args.dataset))
states = pd_states['STATE'].unique()
del pd_states
errors = []
for state in states:
filter = ('STATE', state)
data_pub, _ = randomKway(args.dataset,
args.workload,
args.marginal,
seed=args.workload_seed,
proj=proj,
filter=filter)
data_pub, A_init = get_pub_dataset(data_pub, 1.0, 0)
query_manager.setup_xy(data_pub)
fake_answers = query_manager.get_answer_weights(A_init)
error = np.abs(real_answers - fake_answers)
max_error = error.max()
avg_error = error.mean()
errors.append({
'state': state,
'max_error': max_error,
'avg_error': avg_error
})
print(state, max_error, avg_error)
df = pd.DataFrame(errors).sort_values('max_error').reset_index(drop=True)
df['workload'] = args.workload
df['marginal'] = args.marginal