def run():
results_dir = 'data/results/'
results_dir_input = results_dir + 'input/raw/'
results_dir_output = results_dir + 'input/prepared/openml/'
ag_results_distilled_1h = preprocess_openml.preprocess_openml_input(
path=results_dir_input + 'results_ag_leaderboard_1h_v15_distill.csv',
framework_suffix='_1h')
ag_results_distilled_4h = preprocess_openml.preprocess_openml_input(
path=results_dir_input + 'results_ag_leaderboard_4h_v15_distill.csv',
framework_suffix='_4h')
results_list = [
ag_results_distilled_1h,
ag_results_distilled_4h,
]
results_raw = pd.concat(results_list, ignore_index=True, sort=True)
frameworks_distilled = [
'autogluon_compressed_1h',
'autogluon_distilled_1h',
'autogluon_ensemble_1h',
'autogluon_compressed_4h',
'autogluon_distilled_4h',
'autogluon_ensemble_4h',
]
results_ablation = results_raw[results_raw[FRAMEWORK].isin(
frameworks_distilled)]
save_pd.save(path=results_dir_output + 'openml_autogluon_distilled.csv',
df=results_ablation)
def run():
results_dir = 'data/results/'
results_dir_input = results_dir + 'input/raw/original/'
results_dir_output = results_dir + 'input/prepared/openml/'
other_results_large_4h = preprocess_openml.preprocess_openml_input(path=results_dir_input + 'results_large-8c4h.csv', framework_suffix='_4h')
other_results_medium_4h = preprocess_openml.preprocess_openml_input(path=results_dir_input + 'results_medium-8c4h.csv', framework_suffix='_4h')
other_results_small_4h = preprocess_openml.preprocess_openml_input(path=results_dir_input + 'results_small-8c4h.csv', framework_suffix='_4h')
other_results_medium_1h = preprocess_openml.preprocess_openml_input(path=results_dir_input + 'results_medium-8c1h.csv', framework_suffix='_1h')
other_results_small_1h = preprocess_openml.preprocess_openml_input(path=results_dir_input + 'results_small-8c1h.csv', framework_suffix='_1h')
results_list = [other_results_large_4h, other_results_medium_4h, other_results_small_4h, other_results_medium_1h, other_results_small_1h]
results_raw = pd.concat(results_list, ignore_index=True, sort=True)
results_raw[FRAMEWORK] = ['orig_' + name[0] for name in zip(results_raw[FRAMEWORK])]
frameworks_original = [
'orig_H2OAutoML_1h',
'orig_autosklearn_1h',
'orig_TPOT_1h',
'orig_AutoWEKA_1h',
'orig_H2OAutoML_4h',
'orig_autosklearn_4h',
'orig_TPOT_4h',
'orig_AutoWEKA_4h',
]
results_original = results_raw[results_raw[FRAMEWORK].isin(frameworks_original)]
save_pd.save(path=results_dir_output + 'openml_original.csv', df=results_original)
def run():
results_dir = 'data/results/'
results_dir_input = results_dir + 'input/raw/'
results_dir_output = results_dir + 'input/prepared/openml/'
ag_results_ablation_1h = preprocess_openml.preprocess_openml_input(
path=results_dir_input + 'results_automlbenchmark_ablation_1h.csv',
framework_suffix='_1h')
ag_results_ablation_4h = preprocess_openml.preprocess_openml_input(
path=results_dir_input + 'results_automlbenchmark_ablation_4h.csv',
framework_suffix='_4h')
results_list = [
ag_results_ablation_1h,
ag_results_ablation_4h,
]
results_raw = pd.concat(results_list, ignore_index=True, sort=True)
frameworks_ablation = [
'autogluon_nostack_1h',
'autogluon_nobag_1h',
'autogluon_norepeatbag_1h',
'autogluon_nonn_1h',
'autogluon_noknn_1h',
'autogluon_nostack_4h',
'autogluon_nobag_4h',
'autogluon_norepeatbag_4h',
'autogluon_nonn_4h',
'autogluon_noknn_4h',
]
results_ablation = results_raw[results_raw[FRAMEWORK].isin(
frameworks_ablation)]
save_pd.save(path=results_dir_output + 'openml_autogluon_ablation.csv',
df=results_ablation)
def run():
results_dir = 'data/results/'
results_dir_input = results_dir + 'input/raw/'
results_dir_output = results_dir + 'input/prepared/openml/'
ag_results_1h = preprocess_openml.preprocess_openml_input(
path=results_dir_input + 'results_automlbenchmark_1h.csv',
framework_suffix='_1h')
ag_results_4h = preprocess_openml.preprocess_openml_input(
path=results_dir_input + 'results_automlbenchmark_4h.csv',
framework_suffix='_4h')
ag_results_1h[FRAMEWORK] = ag_results_1h[FRAMEWORK].str.replace(
'_benchmark_', '_', regex=False)
ag_results_4h[FRAMEWORK] = ag_results_4h[FRAMEWORK].str.replace(
'_benchmark_', '_', regex=False)
gcp_results_1h = preprocess_openml.preprocess_openml_input(
path=results_dir_input + 'results_automlbenchmark_gcptables_1h.csv',
framework_suffix='_1h')
gcp_results_4h = preprocess_openml.preprocess_openml_input(
path=results_dir_input + 'results_automlbenchmark_gcptables_4h.csv',
framework_suffix='_4h')
gcp_results_1h[FRAMEWORK] = gcp_results_1h[FRAMEWORK].str.replace(
'GoogleAutoMLTables_benchmark_', 'GCPTables_', regex=False)
gcp_results_4h[FRAMEWORK] = gcp_results_4h[FRAMEWORK].str.replace(
'GoogleAutoMLTables_benchmark_', 'GCPTables_', regex=False)
gcp_results_1h = gcp_results_1h[gcp_results_1h[FRAMEWORK] ==
'GCPTables_1h']
gcp_results_4h = gcp_results_4h[gcp_results_4h[FRAMEWORK] ==
'GCPTables_4h']
results_list = [
gcp_results_1h, gcp_results_4h, ag_results_1h, ag_results_4h
]
results_raw = pd.concat(results_list, ignore_index=True, sort=True)
frameworks_core = [
'autogluon_1h',
'GCPTables_1h',
'H2OAutoML_1h',
'autosklearn_1h',
'TPOT_1h',
'AutoWEKA_1h',
'autogluon_4h',
'GCPTables_4h',
'H2OAutoML_4h',
'autosklearn_4h',
'TPOT_4h',
'AutoWEKA_4h',
]
results_core = results_raw[results_raw[FRAMEWORK].isin(frameworks_core)]
save_pd.save(path=results_dir_output + 'openml_core.csv', df=results_core)
def run():
results_dir = 'data/results/'
results_dir_input = results_dir + 'input/raw/'
results_dir_output = results_dir + 'input/prepared/openml/'
ag_results_autopilot_1h = preprocess_openml.preprocess_openml_input(path=results_dir_input + 'results_automlbenchmark_autopilot_1h.csv', framework_suffix='_1h')
results_list = [ag_results_autopilot_1h]
results_raw = pd.concat(results_list, ignore_index=True, sort=True)
frameworks_autopilot = ['AutoPilot_1h']
results_core = results_raw[results_raw[FRAMEWORK].isin(frameworks_autopilot)]
save_pd.save(path=results_dir_output + 'openml_autopilot.csv', df=results_core)