current_dynamic.append(0.0)
print('dynamic: ' + str(current_dynamic))
print('static: ' + str(current_static))
gen_new = SpaceGenerator()
space_new = gen_new.generate_params()
for pre, _, node in RenderTree(space_new.parameter_tree):
if node.status == True:
print("%s%s" % (pre, node.name))
try:
search = MyAutoML(n_jobs=1,
time_search_budget=search_time_frozen,
space=space_new,
evaluation_budget=int(0.1 *
search_time_frozen),
main_memory_budget_gb=memory_budget,
pipeline_size_limit=pipeline_size,
hold_out_fraction=0.33)
best_result = search.fit(
X_train_hold,
y_train_hold,
categorical_indicator=categorical_indicator_hold,
scorer=my_scorer)
test_score = my_scorer(search.get_best_pipeline(), X_test_hold,
y_test_hold)
except:
test_score = 0.0
current_static.append(test_score)
def run_AutoML(trial, X_train=None, X_test=None, y_train=None, y_test=None, categorical_indicator=None):
space = None
search_time = None
if not 'space' in trial.user_attrs:
# which hyperparameters to use
gen = SpaceGenerator()
space = gen.generate_params()
space.sample_parameters(trial)
trial.set_user_attr('space', copy.deepcopy(space))
search_time, evaluation_time, memory_limit, privacy_limit, training_time_limit, inference_time_limit, pipeline_size_limit, cv, number_of_cvs, hold_out_fraction, sample_fraction, dataset_id = generate_parameters(trial, total_search_time, my_openml_datasets)
else:
space = trial.user_attrs['space']
print(trial.params)
#make this a hyperparameter
search_time = trial.params['global_search_time_constraint']
evaluation_time = search_time
if 'global_evaluation_time_constraint' in trial.params:
evaluation_time = trial.params['global_evaluation_time_constraint']
memory_limit = 10
if 'global_memory_constraint' in trial.params:
memory_limit = trial.params['global_memory_constraint']
privacy_limit = None
if 'privacy_constraint' in trial.params:
privacy_limit = trial.params['privacy_constraint']
training_time_limit = search_time
if 'training_time_constraint' in trial.params:
training_time_limit = trial.params['training_time_constraint']
inference_time_limit = 60
if 'inference_time_constraint' in trial.params:
inference_time_limit = trial.params['inference_time_constraint']
pipeline_size_limit = 350000000
if 'pipeline_size_constraint' in trial.params:
pipeline_size_limit = trial.params['pipeline_size_constraint']
cv = 1
number_of_cvs = 1
hold_out_fraction = None
if 'global_cv' in trial.params:
cv = trial.params['global_cv']
if 'global_number_cv' in trial.params:
number_of_cvs = trial.params['global_number_cv']
else:
hold_out_fraction = trial.params['hold_out_fraction']
sample_fraction = 1.0
if 'sample_fraction' in trial.params:
sample_fraction = trial.params['sample_fraction']
if 'dataset_id' in trial.params:
dataset_id = trial.params['dataset_id']
else:
dataset_id = trial.user_attrs['dataset_id']
for pre, _, node in RenderTree(space.parameter_tree):
if node.status == True:
print("%s%s" % (pre, node.name))
if type(X_train) == type(None):
my_random_seed = int(time.time())
if 'data_random_seed' in trial.user_attrs:
my_random_seed = trial.user_attrs['data_random_seed']
X_train, X_test, y_train, y_test, categorical_indicator, attribute_names = get_data(dataset_id, randomstate=my_random_seed)
if not isinstance(trial, FrozenTrial):
my_list_constraints_values = [search_time,
evaluation_time,
memory_limit, cv,
number_of_cvs,
ifNull(privacy_limit, constant_value=1000),
ifNull(hold_out_fraction),
sample_fraction,
training_time_limit,
inference_time_limit,
pipeline_size_limit]
metafeature_values = data2features(X_train, y_train, categorical_indicator)
features = space2features(space, my_list_constraints_values, metafeature_values)
features = FeatureTransformations().fit(features).transform(features, feature_names=feature_names)
trial.set_user_attr('features', features)
dynamic_params = []
for random_i in range(5): #5
search = MyAutoML(cv=cv,
number_of_cvs=number_of_cvs,
n_jobs=1,
evaluation_budget=evaluation_time,
time_search_budget=search_time,
space=space,
main_memory_budget_gb=memory_limit,
differential_privacy_epsilon=privacy_limit,
hold_out_fraction=hold_out_fraction,
sample_fraction=sample_fraction,
training_time_limit=training_time_limit,
inference_time_limit=inference_time_limit,
pipeline_size_limit=pipeline_size_limit)
test_score = 0.0
try:
search.fit(X_train, y_train, categorical_indicator=categorical_indicator, scorer=my_scorer)
best_pipeline = search.get_best_pipeline()
if type(best_pipeline) != type(None):
test_score = my_scorer(search.get_best_pipeline(), X_test, y_test)
except:
pass
dynamic_params.append(test_score)
count_success = 0
for i_run in range(len(dynamic_params)):
if dynamic_params[i_run] > 0.0:
count_success += 1
success_rate = float(count_success) / float(len(dynamic_params))
return success_rate, search
for pre, _, node in RenderTree(space.parameter_tree):
print("%s%s: %s" % (pre, node.name, node.status))
my_study = optuna.create_study(direction='maximize')
validation_scores = []
test_scores = []
#add Caruana ensemble with replacement # save pipelines to disk
for i in range(1, 10):
search = MyAutoML(cv=10,
number_of_cvs=1,
n_jobs=2,
time_search_budget=2 * 60,
space=space,
study=my_study,
main_memory_budget_gb=4)
best_result = search.fit(X_train,
y_train,
categorical_indicator=categorical_indicator,
scorer=auc)
my_study = search.study
test_score = auc(search.get_best_pipeline(), X_test, y_test)
print("budget: " + str(i) + ' => ' + str(best_result) + " test: " +
str(test_score))
validation_scores.append(best_result)
def run_AutoML(trial, X_train=None, X_test=None, y_train=None, y_test=None, categorical_indicator=None):
space = None
search_time = None
if not 'space' in trial.user_attrs:
# which hyperparameters to use
gen = SpaceGenerator()
space = gen.generate_params()
space.sample_parameters(trial)
trial.set_user_attr('space', copy.deepcopy(space))
# which constraints to use
search_time = trial.suggest_int('global_search_time_constraint', 10, total_search_time, log=False)
# how much time for each evaluation
evaluation_time = trial.suggest_int('global_evaluation_time_constraint', 10, search_time, log=False)
# how much memory is allowed
memory_limit = trial.suggest_uniform('global_memory_constraint', 1.5, 4)
# how many cvs should be used
cv = trial.suggest_int('global_cv', 2, 20, log=False) #todo: calculate minimum number of splits based on y
number_of_cvs = trial.suggest_int('global_number_cv', 1, 10, log=False)
dataset_id = trial.suggest_categorical('dataset_id', my_openml_datasets)
else:
space = trial.user_attrs['space']
print(trial.params)
#make this a hyperparameter
search_time = trial.params['global_search_time_constraint']
evaluation_time = trial.params['global_evaluation_time_constraint']
memory_limit = trial.params['global_memory_constraint']
cv = trial.params['global_cv']
number_of_cvs = trial.params['global_number_cv']
if 'dataset_id' in trial.params:
dataset_id = trial.params['dataset_id'] #get same random seed
else:
dataset_id = 31
for pre, _, node in RenderTree(space.parameter_tree):
print("%s%s: %s" % (pre, node.name, node.status))
# which dataset to use
#todo: add more datasets
if type(X_train) == type(None):
X_train, X_test, y_train, y_test, categorical_indicator, attribute_names = get_data(dataset_id, randomstate=int(time.time()))
if not isinstance(trial, FrozenTrial):
my_list_constraints_values = [search_time, evaluation_time, memory_limit, cv, number_of_cvs]
metafeature_values = data2features(X_train, y_train, categorical_indicator)
features = space2features(space, my_list_constraints_values, metafeature_values)
trial.set_user_attr('features', features)
search = MyAutoML(cv=cv,
number_of_cvs=number_of_cvs,
n_jobs=1,
evaluation_budget=evaluation_time,
time_search_budget=search_time,
space=space,
main_memory_budget_gb=memory_limit)
search.fit(X_train, y_train, categorical_indicator=categorical_indicator, scorer=auc)
best_pipeline = search.get_best_pipeline()
test_score = 0.0
if type(best_pipeline) != type(None):
test_score = auc(search.get_best_pipeline(), X_test, y_test)
return test_score
def utils_run_AutoML(trial,
X_train=None,
X_test=None,
y_train=None,
y_test=None,
categorical_indicator=None,
my_scorer=None,
search_time=None,
memory_limit=None,
privacy_limit=None,
training_time_limit=None,
inference_time_limit=None,
pipeline_size_limit=None):
space = trial.user_attrs['space']
print(trial.params)
if 'evaluation_time' in trial.user_attrs:
evaluation_time = trial.user_attrs['evaluation_time']
else:
evaluation_time = search_time
if 'global_evaluation_time_constraint' in trial.params:
evaluation_time = trial.params['global_evaluation_time_constraint']
cv = 1
number_of_cvs = 1
if 'hold_out_fraction' in trial.user_attrs:
hold_out_fraction = trial.user_attrs['hold_out_fraction']
else:
hold_out_fraction = None
if 'global_cv' in trial.params:
cv = trial.params['global_cv']
if 'global_number_cv' in trial.params:
number_of_cvs = trial.params['global_number_cv']
if 'hold_out_fraction' in trial.params:
hold_out_fraction = trial.params['hold_out_fraction']
sample_fraction = 1.0
if 'sample_fraction' in trial.params:
sample_fraction = trial.params['sample_fraction']
search = MyAutoML(cv=cv,
number_of_cvs=number_of_cvs,
n_jobs=1,
evaluation_budget=evaluation_time,
time_search_budget=search_time,
space=space,
main_memory_budget_gb=memory_limit,
differential_privacy_epsilon=privacy_limit,
hold_out_fraction=hold_out_fraction,
sample_fraction=sample_fraction,
training_time_limit=training_time_limit,
inference_time_limit=inference_time_limit,
pipeline_size_limit=pipeline_size_limit)
search.fit(X_train,
y_train,
categorical_indicator=categorical_indicator,
scorer=my_scorer)
best_pipeline = search.get_best_pipeline()
test_score = 0.0
if type(best_pipeline) != type(None):
test_score = my_scorer(search.get_best_pipeline(), X_test, y_test)
return test_score, search
def run_AutoML(trial, X_train=None, X_test=None, y_train=None, y_test=None, categorical_indicator=None):
space = None
search_time = None
if not 'space' in trial.user_attrs:
# which hyperparameters to use
gen = SpaceGenerator()
space = gen.generate_params()
space.sample_parameters(trial)
trial.set_user_attr('space', copy.deepcopy(space))
search_time, evaluation_time, memory_limit, privacy_limit, cv, number_of_cvs, hold_out_fraction, sample_fraction, dataset_id = generate_parameters(trial)
else:
space = trial.user_attrs['space']
print(trial.params)
#make this a hyperparameter
search_time = total_search_time
evaluation_time = search_time
memory_limit = 4
privacy_limit = None
cv = 1
number_of_cvs = 1
hold_out_fraction = None
if 'global_cv' in trial.params:
cv = trial.params['global_cv']
if 'global_number_cv' in trial.params:
number_of_cvs = trial.params['global_number_cv']
else:
hold_out_fraction = trial.params['hold_out_fraction']
sample_fraction = 1.0
if 'sample_fraction' in trial.params:
sample_fraction = trial.params['sample_fraction']
if 'dataset_id' in trial.params:
dataset_id = trial.params['dataset_id'] #get same random seed
else:
dataset_id = 31
for pre, _, node in RenderTree(space.parameter_tree):
if node.status == True:
print("%s%s" % (pre, node.name))
if type(X_train) == type(None):
my_random_seed = int(time.time())
if 'data_random_seed' in trial.user_attrs:
my_random_seed = trial.user_attrs['data_random_seed']
X_train, X_test, y_train, y_test, categorical_indicator, attribute_names = get_data(dataset_id, randomstate=my_random_seed)
if not isinstance(trial, FrozenTrial):
my_list_constraints_values = [search_time,
evaluation_time,
memory_limit, cv,
number_of_cvs,
ifNull(privacy_limit, constant_value=1000),
ifNull(hold_out_fraction),
sample_fraction]
metafeature_values = data2features(X_train, y_train, categorical_indicator)
features = space2features(space, my_list_constraints_values, metafeature_values)
features = FeatureTransformations().fit(features).transform(features, feature_names=feature_names)
trial.set_user_attr('features', features)
search = MyAutoML(cv=cv,
number_of_cvs=number_of_cvs,
n_jobs=1,
evaluation_budget=evaluation_time,
time_search_budget=search_time,
space=space,
main_memory_budget_gb=memory_limit,
differential_privacy_epsilon=privacy_limit,
hold_out_fraction=hold_out_fraction,
sample_fraction=sample_fraction)
search.fit(X_train, y_train, categorical_indicator=categorical_indicator, scorer=my_scorer)
best_pipeline = search.get_best_pipeline()
test_score = 0.0
if type(best_pipeline) != type(None):
test_score = my_scorer(search.get_best_pipeline(), X_test, y_test)
return test_score, search