def check(in_file='.dat'):
res = load_data(in_file)
print(res)
best_res, mid_res = res
best1, outs1 = mid_res
best_avg_auc = -1
for out in outs1[0]:
if np.mean(out['auc']) > best_avg_auc:
best_avg_auc = np.mean(out['auc'])
best_results = copy.deepcopy(out)
print('\n', best_avg_auc, best_results)
return out
def check(in_file='.dat'):
res = load_data(in_file)
# print(res)
best_res, mid_res = res
# best1: (file_name),
# outs1: [n_repeats][all_middle_res(tuning res)]
best1, outs1 = mid_res
best_avg_auc = -1
for out in outs1[0]:
if np.mean(out['auc']) > best_avg_auc:
best_avg_auc = np.mean(out['auc'])
best_results = copy.deepcopy(out)
print('\n', best_avg_auc, best_results)
return outs1
def dat2xlxs(in_file, out_file):
results = load_data(in_file)
new_results = {}
# methods_mapping = {
# 'detector_name_OCSVM-gs_True-kjl_False-quickshift_False-meanshift_False-nystrom_False': 'OCSVM',
# # 'detector_name_GMM-covariance_type_full-gs_True-kjl_False-meanshift_False-quickshift_False-nystrom_False':'GMM-full',
# # 'detector_name_GMM-covariance_type_diag-gs_True-kjl_False-meanshift_False-quickshift_False-nystrom_False':'GMM-diag',
# # 'detector_name_GMM-covariance_type_full-gs_True-kjl_True-meanshift_False-quickshift_False-nystrom_False':'GMM-full-kjl',
# # 'detector_name_GMM-covariance_type_diag-gs_True-kjl_True-meanshift_False-quickshift_False-nystrom_False':'GMM-diag-kjl',
# # 'detector_name_GMM-covariance_type_full-gs_True-kjl_False-meanshift_False-quickshift_False-nystrom_True': 'GMM-full-nystrom',
# # 'detector_name_GMM-covariance_type_diag-gs_True-kjl_False-meanshift_False-quickshift_False-nystrom_True': 'GMM-diag-nystrom',
# # 'detector_name_GMM-covariance_type_full-gs_True-kjl_True-meanshift_False-quickshift_True-nystrom_False': 'GMM-full-kjl-quickshift',
# # 'detector_name_GMM-covariance_type_diag-gs_True-kjl_True-meanshift_False-quickshift_True-nystrom_False': 'GMM-diag-kjl-quickshift',
# # 'detector_name_GMM-covariance_type_full-gs_True-kjl_True-meanshift_True-quickshift_False-nystrom_False': 'GMM-full-kjl-meanshift',
# # 'detector_name_GMM-covariance_type_diag-gs_True-kjl_True-meanshift_True-quickshift_False-nystrom_False': 'GMM-diag-kjl-meanshift',
#
# 'detector_name_GMM-covariance_type_full-gs_True-kjl_False-nystrom_False-quickshift_False-meanshift_False': 'GMM-full',
# 'detector_name_GMM-covariance_type_diag-gs_True-kjl_False-nystrom_False-quickshift_False-meanshift_False': 'GMM-diag',
# 'detector_name_GMM-covariance_type_full-gs_True-kjl_True-quickshift_False-meanshift_False-nystrom_False': 'GMM-full-kjl',
# 'detector_name_GMM-covariance_type_diag-gs_True-kjl_True-quickshift_False-meanshift_False-nystrom_False': 'GMM-diag-kjl',
# 'detector_name_GMM-covariance_type_full-gs_True-kjl_False-nystrom_True-quickshift_False-meanshift_False': 'GMM-full-nystrom',
# 'detector_name_GMM-covariance_type_diag-gs_True-kjl_False-nystrom_True-quickshift_False-meanshift_False': 'GMM-diag-nystrom',
# 'detector_name_GMM-covariance_type_full-gs_True-kjl_True-meanshift_False-quickshift_True-nystrom_False': 'GMM-full-kjl-quickshift',
# 'detector_name_GMM-covariance_type_diag-gs_True-kjl_True-meanshift_False-quickshift_True-nystrom_False': 'GMM-diag-kjl-quickshift',
# 'detector_name_GMM-covariance_type_full-gs_True-kjl_True-meanshift_True-quickshift_False-nystrom_False': 'GMM-full-kjl-meanshift',
# 'detector_name_GMM-covariance_type_diag-gs_True-kjl_True-meanshift_True-quickshift_False-nystrom_False': 'GMM-diag-kjl-meanshift',
#
# }
GMM_covariance_type = 'diag'
methods_mapping = {
'case1': 'OCSVM',
'case2': f'GMM-{GMM_covariance_type}',
'case3': f'GMM-{GMM_covariance_type}-kjl',
'case4': f'GMM-{GMM_covariance_type}-nystrom',
'case5': f'GMM-{GMM_covariance_type}-kjl-quickshift',
'case6': f'GMM-{GMM_covariance_type}-kjl-meanshift',
}
with ExcelWriter(out_file) as writer:
for sheet_name in ['OCSVM']:
values = []
for i, ((data_key, method_key), (best_values, mid_values)) in enumerate(results.items()):
if methods_mapping[method_key] != sheet_name:
continue
print(f'\n\n----{i}, {sheet_name}')
try:
header = 5
X_train_shape = best_values['X_train_shape']
X_test_shape = best_values['X_test_shape']
aucs = best_values['aucs']
train_times = best_values['train_times']
test_times = best_values['test_times']
aucs_str = "-".join([str(v) for v in aucs])
train_times_str = "-".join([str(v) for v in train_times])
test_times_str = "-".join([str(v) for v in test_times])
params = best_values['params']
_suffex = ''
line = f'{data_key}, {X_train_shape}, {X_test_shape}, => aucs:{aucs_str}, train_times:{train_times_str}, test_times:{test_times_str}, with params: {params}: {_suffex}'
values.append(line)
pd.DataFrame(values).to_excel(writer, sheet_name, index=False, header=False)
startrow = len(values) + 5
_each_total_rows = len(values) + 5
except Exception as e:
traceback.print_exc()
values = [sheet_name]
values.extend([] * _each_total_rows)
writer.save()
return out_file
def dat2xlxs_new(in_file, out_file, models=None):
results = load_data(in_file)
try:
for model_name, vs in results.items():
if 'GMM' in model_name:
if 'full' in model_name:
GMM_covariance_type = 'full'
elif 'diag' in model_name:
GMM_covariance_type = 'diag'
else:
msg = model_name
raise NotImplementedError(msg)
except Exception as e:
print(f'e: {e}')
GMM_covariance_type = 'diag'
with ExcelWriter(out_file) as writer:
for i, sheet_name in enumerate(models):
values = []
_each_total_rows = 0
is_file_header = True
if sheet_name not in results.keys():
continue
res_ = results[sheet_name]
for j, vs in enumerate(res_):
print(
f'----i={i}, j={j}, sheet_name:{sheet_name}, model_name: {sheet_name}, {vs}'
)
try:
if is_file_header:
header = 5
# the first 5 rows are empty for writing configurations
values.extend([[] for i in range(header)])
values.append([sheet_name])
is_file_header = False
data_key = vs[0]
try:
X_train_shape = vs[2].split(
'X_train_shape:')[-1].split('|')[0]
X_train_shape = f'{X_train_shape}'.replace(',', '-')
X_val_shape = vs[2].split('X_val_shape:')[-1]
X_val_shape = f'{X_val_shape}'.replace(',', '-')
X_test_shape = vs[3].split('X_test_shape:')[-1]
X_test_shape = f'{X_test_shape}'.replace(',', '-')
aucs = vs[7].split(":")[-1]
train_times = vs[8].split(":")[-1]
test_times = vs[9].split(":")[-1]
space_sizes = vs[12].split(":")[-1]
# aucs_str = "-".join([str(v) for v in aucs])
# train_times_str = "-".join([str(v) for v in train_times])
# test_times_str = "-".join([str(v) for v in test_times])
# space_sizes_str = "-".join([str(v) for v in space_sizes])
params = vs[13:]
# n_comps = [int(v['GMM_n_components']) for v in best_values['params']]
# mu_n_comp = np.mean(n_comps)
# std_n_comp = np.std(n_comps)
# n_comp_str = '-'.join([str(v) for v in n_comps])
# n_comp_str = f'{n_comp_str}|(u_std) {mu_n_comp:.2f}+/-{std_n_comp:.2f}'
n_comp_str = vs[10].split(":")[-1]
_suffex = f', n_comps_arr: {n_comp_str}'
line = f'{data_key}, {sheet_name}, {X_train_shape}|{X_val_shape}, {X_test_shape}, ' \
f'auc: u+/-std, , , ' \
f'=> aucs:{aucs}, train_times:{train_times}, test_times:{test_times}, ' \
f'space_sizes: {space_sizes}, with params: {params}: {_suffex}'
except Exception as e:
# n_comp_str = f'error-{e}'
# _suffex = f', n_comps_arr: {n_comp_str}'
line = f'{data_key}, {sheet_name}, X_train_shape|X_val_shape, X_test_shape, ' \
f'auc: u+/-std, , , ' \
f'=> aucs:, train_times:, test_times:, ' \
f'space_sizes: , with params: : '
values.append(line.split(','))
pd.DataFrame(values).to_excel(writer,
sheet_name,
index=False,
header=False)
startrow = len(values) + 5
_each_total_rows = len(values) + 5
except Exception as e:
traceback.print_exc()
values = [sheet_name]
values.extend([] * _each_total_rows)
# once call write.save(), it will close the file
writer.save()
return out_file
def dat2xlxs(
in_file,
out_file,
):
results = load_data(in_file)
new_results = {}
# methods_mapping = {
# 'detector_name_OCSVM-gs_True-kjl_False-quickshift_False-meanshift_False-nystrom_False': 'OCSVM',
# # 'detector_name_GMM-covariance_type_full-gs_True-kjl_False-meanshift_False-quickshift_False-nystrom_False':'GMM-full',
# # 'detector_name_GMM-covariance_type_diag-gs_True-kjl_False-meanshift_False-quickshift_False-nystrom_False':'GMM-diag',
# # 'detector_name_GMM-covariance_type_full-gs_True-kjl_True-meanshift_False-quickshift_False-nystrom_False':'GMM-full-kjl',
# # 'detector_name_GMM-covariance_type_diag-gs_True-kjl_True-meanshift_False-quickshift_False-nystrom_False':'GMM-diag-kjl',
# # 'detector_name_GMM-covariance_type_full-gs_True-kjl_False-meanshift_False-quickshift_False-nystrom_True': 'GMM-full-nystrom',
# # 'detector_name_GMM-covariance_type_diag-gs_True-kjl_False-meanshift_False-quickshift_False-nystrom_True': 'GMM-diag-nystrom',
# # 'detector_name_GMM-covariance_type_full-gs_True-kjl_True-meanshift_False-quickshift_True-nystrom_False': 'GMM-full-kjl-quickshift',
# # 'detector_name_GMM-covariance_type_diag-gs_True-kjl_True-meanshift_False-quickshift_True-nystrom_False': 'GMM-diag-kjl-quickshift',
# # 'detector_name_GMM-covariance_type_full-gs_True-kjl_True-meanshift_True-quickshift_False-nystrom_False': 'GMM-full-kjl-meanshift',
# # 'detector_name_GMM-covariance_type_diag-gs_True-kjl_True-meanshift_True-quickshift_False-nystrom_False': 'GMM-diag-kjl-meanshift',
#
# 'detector_name_GMM-covariance_type_full-gs_True-kjl_False-nystrom_False-quickshift_False-meanshift_False': 'GMM-full',
# 'detector_name_GMM-covariance_type_diag-gs_True-kjl_False-nystrom_False-quickshift_False-meanshift_False': 'GMM-diag',
# 'detector_name_GMM-covariance_type_full-gs_True-kjl_True-quickshift_False-meanshift_False-nystrom_False': 'GMM-full-kjl',
# 'detector_name_GMM-covariance_type_diag-gs_True-kjl_True-quickshift_False-meanshift_False-nystrom_False': 'GMM-diag-kjl',
# 'detector_name_GMM-covariance_type_full-gs_True-kjl_False-nystrom_True-quickshift_False-meanshift_False': 'GMM-full-nystrom',
# 'detector_name_GMM-covariance_type_diag-gs_True-kjl_False-nystrom_True-quickshift_False-meanshift_False': 'GMM-diag-nystrom',
# 'detector_name_GMM-covariance_type_full-gs_True-kjl_True-meanshift_False-quickshift_True-nystrom_False': 'GMM-full-kjl-quickshift',
# 'detector_name_GMM-covariance_type_diag-gs_True-kjl_True-meanshift_False-quickshift_True-nystrom_False': 'GMM-diag-kjl-quickshift',
# 'detector_name_GMM-covariance_type_full-gs_True-kjl_True-meanshift_True-quickshift_False-nystrom_False': 'GMM-full-kjl-meanshift',
# 'detector_name_GMM-covariance_type_diag-gs_True-kjl_True-meanshift_True-quickshift_False-nystrom_False': 'GMM-diag-kjl-meanshift',
#
# }
try:
GMM_covariance_type = \
[_best_res['params'][-1]['GMM_covariance_type'] for (data_name_file, case), (_best_res, _mid_res) in
results.items() if 'GMM_covariance_type' in _best_res['params'][-1].keys()][0]
except Exception as e:
print(f'e: {e}')
GMM_covariance_type = 'diag'
methods_mapping = {
'case1': 'OCSVM(rbf)',
'case1_2': 'KJL-OCSVM(linear)',
'case1_3': 'Nystrom-OCSVM(linear)',
'case2': f'GMM({GMM_covariance_type})',
'case3': f'KJL-GMM({GMM_covariance_type})',
'case4': f'Nystrom-GMM({GMM_covariance_type})',
'case5': f'KJL-QS-GMM({GMM_covariance_type})',
'case6': f'KJL-MS-GMM({GMM_covariance_type})',
'case7': f'Nystrom-QS-GMM({GMM_covariance_type})',
'case8': f'Nystrom-MS-GMM({GMM_covariance_type})',
}
methods_mapping = OrderedDict(methods_mapping)
with ExcelWriter(out_file) as writer:
for i, sheet_name in enumerate(methods_mapping.values()):
values = []
_each_total_rows = 0
is_header = True
for j, ((data_key, method_key),
(best_values, mid_values)) in enumerate(results.items()):
if methods_mapping[method_key] != sheet_name:
continue
print(
f'----i={i}, j={j}, sheet_name:{sheet_name}, data_key: {data_key}'
)
try:
if is_header:
header = 5
# the first 5 rows are empty for writing configurations
values.extend([[] for i in range(header)])
values.append([method_key])
is_header = False
X_train_shape = best_values['X_train_shape']
X_train_shape = f'{X_train_shape}'.replace(',', '-')
try:
X_val_shape = best_values['X_val_shape']
X_val_shape = f'{X_val_shape}'.replace(',', '-')
except Exception as e:
X_val_shape = f'(-)'
X_test_shape = best_values['X_test_shape']
X_test_shape = f'{X_test_shape}'.replace(',', '-')
aucs = best_values['aucs']
train_times = best_values['train_times']
test_times = best_values['test_times']
space_sizes = best_values['space_sizes']
aucs_str = "-".join([str(v) for v in aucs])
train_times_str = "-".join([str(v) for v in train_times])
test_times_str = "-".join([str(v) for v in test_times])
space_sizes_str = "-".join([str(v) for v in space_sizes])
params = best_values['params'][-1]
try:
n_comps = [
int(v['GMM_n_components'])
for v in best_values['params']
]
mu_n_comp = np.mean(n_comps)
std_n_comp = np.std(n_comps)
n_comp_str = '-'.join([str(v) for v in n_comps])
n_comp_str = f'{n_comp_str}|(u_std) {mu_n_comp:.2f}+/-{std_n_comp:.2f}'
except Exception as e:
n_comp_str = f'error-{e}'
_suffex = f', n_comps_arr: {n_comp_str}'
line = f'{data_key}, {method_key}, {X_train_shape}|{X_val_shape}, {X_test_shape}, ' \
f'auc: u+/-std, , , ' \
f'=> aucs:{aucs_str}, train_times:{train_times_str}, test_times:{test_times_str}, ' \
f'space_sizes: {space_sizes_str}, with params: {params}: {_suffex}'
values.append(line.split(','))
pd.DataFrame(values).to_excel(writer,
sheet_name,
index=False,
header=False)
startrow = len(values) + 5
_each_total_rows = len(values) + 5
except Exception as e:
traceback.print_exc()
values = [sheet_name]
values.extend([] * _each_total_rows)
# once call write.save(), it will close the file
writer.save()
return out_file
def main():
data_mapping = {
# 'DS10_UNB_IDS/DS11-srcIP_192.168.10.5':'UNB_PC1',
'DS10_UNB_IDS/DS12-srcIP_192.168.10.8': 'UNB_PC2',
'DS10_UNB_IDS/DS13-srcIP_192.168.10.9': 'UNB_PC3',
'DS10_UNB_IDS/DS14-srcIP_192.168.10.14': 'UNB_PC4',
'DS10_UNB_IDS/DS15-srcIP_192.168.10.15': 'UNB_PC5',
# 'DS20_PU_SMTV/DS21-srcIP_10.42.0.1':'SMTV',
# # #
'DS40_CTU_IoT/DS41-srcIP_10.0.2.15': 'CTU',
'DS40_CTU_IoT/DS42-srcIP_192.168.1.196': 'CTU',
#
# 'DS50_MAWI_WIDE/DS51-srcIP_202.171.168.50':'MAWI',
# 'DS50_MAWI_WIDE/DS51-srcIP_202.171.168.50',
'DS50_MAWI_WIDE/DS52-srcIP_203.78.7.165': 'MAWI_PC2',
# 'DS50_MAWI_WIDE/DS53-srcIP_203.78.4.32':'MAWI_PC3',
# 'DS50_MAWI_WIDE/DS54-srcIP_222.117.214.171':'MAWI_PC4',
# 'DS50_MAWI_WIDE/DS55-srcIP_101.27.14.204':'MAWI_PC5',
# 'DS50_MAWI_WIDE/DS56-srcIP_18.178.219.109':'MAWI_PC6',
'ISTS/2015': 'ISTS',
'MACCDC/2012': 'MACCDC',
# 'DS60_UChi_IoT/DS61-srcIP_192.168.143.20':'GHom',
'DS60_UChi_IoT/DS62-srcIP_192.168.143.42': 'SCam',
# 'DS60_UChi_IoT/DS63-srcIP_192.168.143.43':'SFrig',
# 'DS60_UChi_IoT/DS64-srcIP_192.168.143.48':'BSTCH',
}
datasets = [
# # # # 'DS10_UNB_IDS/DS11-srcIP_192.168.10.5',
'DS10_UNB_IDS/DS12-srcIP_192.168.10.8',
# # 'DS10_UNB_IDS/DS13-srcIP_192.168.10.9',
# # 'DS10_UNB_IDS/DS14-srcIP_192.168.10.14',
# # 'DS10_UNB_IDS/DS15-srcIP_192.168.10.15',
# # # # # #
# # # # # 'DS20_PU_SMTV/DS21-srcIP_10.42.0.1',
# # # # # # #
# 'DS40_CTU_IoT/DS41-srcIP_10.0.2.15',
'DS40_CTU_IoT/DS42-srcIP_192.168.1.196',
# # # #
# # # # # 'DS50_MAWI_WIDE/DS51-srcIP_202.171.168.50',
# # 'DS50_MAWI_WIDE/DS51-srcIP_202.171.168.50',
'DS50_MAWI_WIDE/DS52-srcIP_203.78.7.165',
# # 'DS50_MAWI_WIDE/DS53-srcIP_203.78.4.32',
# # 'DS50_MAWI_WIDE/DS54-srcIP_222.117.214.171',
# # 'DS50_MAWI_WIDE/DS55-srcIP_101.27.14.204',
# # 'DS50_MAWI_WIDE/DS56-srcIP_18.178.219.109',
# # #
# # # #
# # 'WRCCDC/2020-03-20',
# # 'DEFCON/ctf26',
'ISTS/2015',
'MACCDC/2012',
# # # # # 'DS60_UChi_IoT/DS61-srcIP_192.168.143.20',
'DS60_UChi_IoT/DS62-srcIP_192.168.143.42',
# # # # 'DS60_UChi_IoT/DS63-srcIP_192.168.143.43',
# # # # 'DS60_UChi_IoT/DS64-srcIP_192.168.143.48'
#
]
# dataset_name = datasets[0]
for dataset_name in datasets:
print(f'\n\n***dataset_name: {dataset_name}')
# data_type = 'all_results'
data_type = 'middle_results'
if data_type == 'all_results':
in_file = 'out/data_kjl/all_results.dat'
results = load_data(in_file)
print(results)
elif data_type == 'middle_results':
if 'DS10_UNB_IDS/DS12-srcIP_192.168.10.8' in dataset_name:
in_file = f'out/data_kjl/{dataset_name}/iat_size/header:False/' \
'detector_name_GMM-covariance_type_full-gs_True-kjl_True-quickshift_False-meanshift_False-nystrom_False.csv-middle_results.dat'
params = {'file_name': dataset_name,
'detector': 'detector_name_GMM-covariance_type_full-gs_True-kjl_True-quickshift_False-meanshift_False-nystrom_False',
'n_components': 5,
'q': 0.4,
'd': 10,
'n': 100
}
elif 'DS40_CTU_IoT/DS42-srcIP_192.168.1.196' in dataset_name:
in_file = f'out/data_kjl/{dataset_name}/iat_size/header:False/' \
'detector_name_GMM-covariance_type_full-gs_True-kjl_True-quickshift_False-meanshift_False-nystrom_False.csv-middle_results.dat'
params = {'file_name': dataset_name,
'detector': 'detector_name_GMM-covariance_type_full-gs_True-kjl_True-quickshift_False-meanshift_False-nystrom_False',
'n_components': 5,
'q': 0.2,
'd': 10,
'n': 100
}
elif 'DS50_MAWI_WIDE/DS52-srcIP_203.78.7.165' in dataset_name:
in_file = f'out/data_kjl/{dataset_name}/iat_size/header:False/' \
'detector_name_GMM-covariance_type_full-gs_True-kjl_True-quickshift_False-meanshift_False-nystrom_False.csv-middle_results.dat'
params = {'file_name': dataset_name,
'detector': 'detector_name_GMM-covariance_type_full-gs_True-kjl_True-quickshift_False-meanshift_False-nystrom_False',
'n_components': 45,
'q': 0.9,
'd': 10,
'n': 100
}
elif 'ISTS/2015' == dataset_name:
in_file = f'out/data_kjl/{dataset_name}/iat_size/header:False/' \
'detector_name_GMM-covariance_type_full-gs_True-kjl_True-quickshift_False-meanshift_False-nystrom_False.csv-middle_results.dat'
params = {'file_name': dataset_name,
'detector': 'detector_name_GMM-covariance_type_full-gs_True-kjl_True-quickshift_False-meanshift_False-nystrom_False',
'n_components': 45,
'q': 0.2,
'd': 10,
'n': 100
}
elif 'MACCDC/2012' == dataset_name:
in_file = f'out/data_kjl/{dataset_name}/iat_size/header:False/' \
'detector_name_GMM-covariance_type_full-gs_True-kjl_True-quickshift_False-meanshift_False-nystrom_False.csv-middle_results.dat'
params = {'file_name': dataset_name,
'detector': 'detector_name_GMM-covariance_type_full-gs_True-kjl_True-quickshift_False-meanshift_False-nystrom_False',
'n_components': 35,
'q': 0.4,
'd': 10,
'n': 100
}
elif 'DS60_UChi_IoT/DS62-srcIP_192.168.143.42' == dataset_name:
in_file = f'out/data_kjl/{dataset_name}/iat_size/header:False/' \
'detector_name_GMM-covariance_type_full-gs_True-kjl_True-quickshift_False-meanshift_False-nystrom_False.csv-middle_results.dat'
params = {'file_name': dataset_name,
'detector': 'detector_name_GMM-covariance_type_full-gs_True-kjl_True-quickshift_False-meanshift_False-nystrom_False',
'n_components': 40,
'q': 0.7,
'd': 10,
'n': 100
}
in_file = 'out/data_kjl/all_results-kjl.dat'
results = load_data(in_file)
# get middle results
results = results[(f'data/data_kjl/{dataset_name}/iat_size/header:False',
'detector_name_GMM-covariance_type_full-gs_True-kjl_True-quickshift_False-meanshift_False-nystrom_False')][
1]
# pprint(results)
kjl_n_aucs, kjl_d_aucs, kjl_q_aucs, n_components_aucs = get_each(results, params)
title = data_mapping[params['file_name']]
y = [np.mean(v) for v in list(kjl_n_aucs.values())]
y_err = [np.std(v) for v in list(kjl_n_aucs.values())]
print(f'{dataset_name}, n (varies): {y_err}')
plot_err_bar(x=list(kjl_n_aucs.keys()), y=y, y_err=y_err, xlabel='n',
ylabel='auc',
title=title + '\n(d={}, q={} (of kjl) and n_components={} (of GMM))'.format(params['d'],
params['q'],
params[
'n_components']))
y = [np.mean(v) for v in list(kjl_d_aucs.values())]
y_err = [np.std(v) for v in list(kjl_d_aucs.values())]
print(f'{dataset_name}, d (varies): {y_err}')
plot_err_bar(x=list(kjl_d_aucs.keys()), y=y, y_err=y_err, xlabel='d',
ylabel='auc',
title=title + '\n(n={}, q={} (of kjl) and n_components={} (of GMM))'.format(params['n'],
params['q'],
params[
'n_components']))