random.seed(r*10)
items = list(df)
items.remove('grid')
items.remove('grid_group')
items.remove('uid')
random.shuffle(items)
items_train = items[:int(len(items) * 0.8)]
items_test = list(set(items) - set(items_train))
df_train = df_cluster.drop(items_test, axis=1)
df_test = df_cluster.drop(items_train, axis=1)
if method == 'HyObscure':
X_obf_dict, X_ori, model_rf, model_xgb = obfuscations.HyObscure(df_train, grid_area_dict, area_grid_dict, cluster_num, grid_area_number, grid_list,
area_reducibility, area_grid_rowcol_dict, area_grid_colrow_dict, method,
grid_rowcol, grid_colrow, l_threshold, k_threshold, deltaX, pp)
elif method == 'YGen':
X_obf_dict, X_ori, model_rf, model_xgb = obfuscations.YGen(df_train, grid_area_dict, grid_area_number, cluster_num, grid_list, area_grid_dict, deltaX,
area_reducibility, area_grid_rowcol_dict, area_grid_colrow_dict, method,
grid_rowcol, grid_colrow, l_threshold, k_threshold, pp)
elif method == 'XObf':
X_obf_dict, X_ori, model_rf, model_xgb = obfuscations.XObf(df_train, cluster_num, grid_area_number, grid_list, grid_area_dict, area_grid_dict, deltaX, pp, method)
elif method == 'PrivCheck':
X_obf_dict, X_ori, model_rf, model_xgb = obfuscations.PrivCheck(df_train, cluster_num, grid_list, grid_area_dict, grid_area_number, area_grid_dict, deltaX, pp)
elif method == 'DP':
X_obf_dict, X_ori, model_rf, model_xgb = obfuscations.differential_privacy(df_train, grid_area_dict, grid_area_number, beta)
elif method == 'Frapp':
X_obf_dict, X_ori, model_rf, model_xgb = obfuscations.Frapp(df_train, grid_area_dict, gamma)
elif method == 'Random':
X_obf_dict, X_ori, model_rf, model_xgb = obfuscations.Random(df_train, grid_area_dict, p_rand)
funcs.update_age_group(df_cluster, age_group_dict)
random.seed(r * 10)
items = list(df_item_age_uid)
items.remove('age')
items.remove('age_group')
items.remove('uid')
random.shuffle(items)
items_train = items[:int(len(items) * 0.8)]
items_test = list(set(items) - set(items_train))
df_train = df_cluster.drop(items_test, axis=1)
df_test = df_cluster.drop(items_train, axis=1)
if method == 'HyObscure':
X_obf_dict, X_ori = obfuscations.HyObscure(
cluster_num, age_group_number, age_group_dict,
group_age_dict, age_list, deltaX, k_threshold, l_threshold,
df_train, df_item_age_uid, pp)
elif method == 'YGen':
X_obf_dict, X_ori = obfuscations.YGen(
df_train, age_group_number, cluster_num, age_list,
age_group_dict, group_age_dict, df_item_age_uid, deltaX,
k_threshold, l_threshold, pp)
elif method == 'XObf':
X_obf_dict, X_ori = obfuscations.XObf(deltaX, cluster_num,
age_group_number,
age_list, group_age_dict,
df_train, pp)
elif method == 'PrivCheck':
X_obf_dict, X_ori = obfuscations.PrivCheck(
deltaX, cluster_num, age_group_number, df_cluster,
df_train, age_list, age_group_dict, group_age_dict, pp)
if cluster_flag == 1:
continue
df_test_rec_items = df_test_items.loc[test_idx_list]
df_test_rec_items = df_test_rec_items.reset_index(drop=True)
print("all users num: {}".format(len(df_cluster)))
print("split train and test over")
print("train num {}".format(len(df_train)))
print("test num {}".format(len(df_test)))
print("train items {}".format(df_train_items.shape[1]))
print("test items {}".format(df_test_items.shape[1]))
if method == 'HyObscure':
X_obf_dict, X_ori, model_rf, model_xgb = obfuscations.HyObscure(
deltaX, grid_area_number, cluster_num, k_threshold,
l_threshold, df_test, grid_list, area_reducibility,
grid_area_dict, area_grid_dict, area_grid_colrow_dict,
area_grid_rowcol_dict, grid_colrow, grid_rowcol, df_train,
df_test_rec_items, pp, method)
elif method == 'PrivCheck':
X_obf_dict, X_ori, model_rf, model_xgb = obfuscations.PrivCheck(
df_train, df_test, df_test_rec_items, grid_area_dict,
area_grid_dict, grid_list, cluster_num, grid_area_number,
deltaX, pp)
else:
print('Method error. Check method setting.')
break
acc_oris_rf = []
acc_obfs_rf = []
acc_oris_xgb = []
acc_obfs_xgb = []
df_test = df_test.reset_index(drop=True)
df_test_rec_items = df_test_items.loc[test_idx_list]
df_test_rec_items = df_test_rec_items.reset_index(drop=True)
print("all users num: {}".format(len(df_item_ageGroup_uid)))
print("split train and test over")
print("train num {}".format(len(df_train)))
print("test num {}".format(len(df_test)))
print("train items {}".format(df_train_items.shape[1]))
print("test items {}".format(df_test_items.shape[1]))
if method == 'HyObscure':
X_obf_dict, X_ori, model_rf, model_xgb = obfuscations.HyObscure(
df_train, df_test, df_test_rec_items, df_item_age_uid,
age_group_dict, group_age_dict, cluster_num,
age_group_number, age_list, deltaX, k_threshold,
l_threshold, pp)
elif method == 'PrivCheck':
X_obf_dict, X_ori, model_rf, model_xgb = obfuscations.PrivCheck(
df_train, df_test, df_test_rec_items, age_group_number,
cluster_num, deltaX, age_list, age_group_dict,
group_age_dict, pp)
else:
print('Method error. Check method setting.')
break
mae_oris_rf = []
mae_obfs_rf = []
mae_oris_xgb = []
mae_obfs_xgb = []
