def test_multiple_run(self):
exp_name = 'TEST_MULTIPLE'
n_runs = 2
self.args += [
'--run-multiple',
str(n_runs), self.testing_datasets[0], exp_name
]
run(args=self.args)
session = local_create_session(self.db_file)
results = session.query(Result).filter(Result.name == exp_name).all()
n_results = len(results)
session.close()
self.assertEqual(n_runs, n_results)
def test_feature_limit_ga(self):
for dataset in self.testing_datasets:
args = self.args + [dataset, self.exp_name, '--max-features', '50']
result_ids = run(args=args)
result_id = result_ids[0]
session = local_create_session(self.db_file)
result: Result = session.query(Result).filter(
Result.id == result_id).first()
features = result.selected_features
session.close()
self.assertLessEqual(len(features), 50)
def test_ga(self):
for dataset in self.testing_datasets:
print(dataset)
args = self.args + [dataset, self.exp_name]
run(args=args)
def test_pso(self):
for dataset in self.testing_datasets:
args = self.args + [dataset, self.exp_name, '--strategy', 'pso']
run(args=args)
def main():
run_multiple = 3
db_file = 'affinityProp.db'
affinity_preferences = {
"('raw',)": {
'campus_basin': -650,
'equatorial_margin': -400,
'talara_basin': -400,
'carmopolisGrouped': -540,
'jequitinhonha': -470,
'mucuri': -5300
},
"('compositional_groups', 'localizational_groups')": {
'campus_basin': -1550,
'equatorial_margin': -900,
'talara_basin': -750,
'carmopolisGrouped': -1590,
'jequitinhonha': -900,
'mucuri': -7700
}
}
# args = {
# '-e': '0.1',
# '--num-gen': '2',
# '--pop-size': '256',
# '--min-features': '2',
# '--strategy': 'none',
# '--db-file': db_file,
# '--cluster-algorithm': 'agglomerative'
# }
args = {
'--strategy': 'none',
'--db-file': db_file,
'--cluster-algorithm': 'affinity-propagation',
}
# args = {
# '--strategy': 'none',
# '--db-file': db_file,
# '--cluster-algorithm': 'kmeans'
# }
datasets_folder = './datasets/'
dataset_locations = {
'campus_basin':
datasets_folder + '/CampusBasin/subtotals_dataset.xlsx',
'equatorial_margin':
datasets_folder + '/MargemEquatorial/subtotals_dataset.xlsx',
'talara_basin':
datasets_folder + '/TalaraBasin/subtotals_dataset.xlsx',
# 'carmopolis': datasets_folder + '/Carmopolis/subtotals_dataset.xlsx',
'carmopolisGrouped':
datasets_folder + '/CarmopolisGrouped/subtotals_dataset.xlsx',
'jequitinhonha':
datasets_folder + '/Jequitinhonha/subtotals_dataset.xlsx',
'mucuri': datasets_folder + '/Mucuri/subtotals_dataset.xlsx'
}
for name, dataset_file in dataset_locations.items():
if not os.path.isfile(dataset_file):
raise FileNotFoundError(f'{dataset_file} not found for {name}')
for scenario in [('raw', ),
('compositional_groups', 'localizational_groups')]:
print(scenario)
local_args = args
local_args['--scenario'] = list(scenario)
fitness_metric = 'silhouette_sklearn'
local_args['--fitness-metric'] = fitness_metric
for experiment_name, dataset_file in dataset_locations.items():
print(f'RUNNING {experiment_name} in scenario {str(scenario)}')
if local_args['--cluster-algorithm'] == 'affinity-propagation':
print(affinity_preferences[str(scenario)][experiment_name])
local_args['--preference'] = str(
affinity_preferences[str(scenario)][experiment_name])
for i in range(run_multiple):
print(f'{i+1}/{run_multiple}')
list_local_args = list(chain.from_iterable(local_args.items()))
run(args=list_local_args + [dataset_file, experiment_name])