def test__handling_alias_parameters():
# type: () -> None
params = {'reg_alpha': 0.1}
_handling_alias_parameters(params)
assert 'reg_alpha' not in params
assert 'lambda_l1' in params
def test__handling_alias_parameters():
# type: () -> None
params = {"reg_alpha": 0.1}
_handling_alias_parameters(params)
assert "reg_alpha" not in params
assert "lambda_l1" in params
def run(self):
# type: () -> lgb.Booster
"""Perform the hyperparameter-tuning with given parameters.
Returns:
booster : Booster
The trained Booster model.
"""
# Surpress log messages.
if self.auto_options['verbosity'] == 0:
optuna.logging.disable_default_handler()
self.lgbm_params['verbose'] = -1
self.lgbm_params['seed'] = 111
self.lgbm_kwargs['verbose_eval'] = False
# Handling aliases.
_handling_alias_parameters(self.lgbm_params)
# Sampling.
self.sample_train_set()
# Tuning.
time_budget = self.auto_options['time_budget']
self.start_time = time.time()
with _timer() as t:
self.tune_feature_fraction()
if time_budget is not None and time_budget > t.elapsed_secs():
self.best_params.update(self._get_params())
return self.best_booster
self.tune_num_leaves()
if time_budget is not None and time_budget > t.elapsed_secs():
self.best_params.update(self._get_params())
return self.best_booster
self.tune_bagging()
if time_budget is not None and time_budget > t.elapsed_secs():
self.best_params.update(self._get_params())
return self.best_booster
self.tune_feature_fraction_stage2()
if time_budget is not None and time_budget > t.elapsed_secs():
self.best_params.update(self._get_params())
return self.best_booster
self.tune_regularization_factors()
if time_budget is not None and time_budget > t.elapsed_secs():
self.best_params.update(self._get_params())
return self.best_booster
self.tune_min_data_in_leaf()
if time_budget is not None and time_budget > t.elapsed_secs():
self.best_params.update(self._get_params())
return self.best_booster
self.best_params.update(self._get_params())
return self.best_booster
def test_handling_alias_parameter():
# type: () -> None
params = {
"num_boost_round": 5,
"early_stopping_rounds": 2,
"min_data": 0.2,
}
_handling_alias_parameters(params)
assert "min_data" not in params
assert "min_data_in_leaf" in params
assert params["min_data_in_leaf"] == 0.2
def test_handling_alias_parameter():
# type: () -> None
params = {
'num_boost_round': 5,
'early_stopping_rounds': 2,
'min_data': 0.2,
}
_handling_alias_parameters(params)
assert 'min_data' not in params
assert 'min_data_in_leaf' in params
assert params['min_data_in_leaf'] == 0.2
def test_handling_alias_parameter_with_default_value():
# type: () -> None
params = {
'num_boost_round': 5,
'early_stopping_rounds': 2,
}
_handling_alias_parameters(params)
assert 'eta' not in params
assert 'learning_rate' in params
assert params['learning_rate'] == 0.1
def test_handling_alias_parameter_with_user_supplied_param():
# type: () -> None
params = {
"num_boost_round": 5,
"early_stopping_rounds": 2,
"eta": 0.5,
}
_handling_alias_parameters(params)
assert "eta" not in params
assert "learning_rate" in params
assert params["learning_rate"] == 0.5
def test_handling_alias_parameter_with_user_supplied_param():
# type: () -> None
params = {
'num_boost_round': 5,
'early_stopping_rounds': 2,
'eta': 0.5,
}
_handling_alias_parameters(params)
assert 'eta' not in params
assert 'learning_rate' in params
assert params['learning_rate'] == 0.5
def run(self) -> None:
"""Perform the hyperparameter-tuning with given parameters."""
# Surpress log messages.
if self.auto_options["verbosity"] == 0:
optuna.logging.disable_default_handler()
self.lgbm_params["verbose"] = -1
self.lgbm_params["seed"] = 111
self.lgbm_kwargs["verbose_eval"] = False
# Handling aliases.
_handling_alias_parameters(self.lgbm_params)
# Sampling.
self.sample_train_set()
# Tuning.
time_budget = self.auto_options["time_budget"]
self.start_time = time.time()
with _timer() as t:
self.tune_feature_fraction()
if time_budget is not None and time_budget < t.elapsed_secs():
return
self.tune_num_leaves()
if time_budget is not None and time_budget < t.elapsed_secs():
return
self.tune_bagging()
if time_budget is not None and time_budget < t.elapsed_secs():
return
self.tune_feature_fraction_stage2()
if time_budget is not None and time_budget < t.elapsed_secs():
return
self.tune_regularization_factors()
if time_budget is not None and time_budget < t.elapsed_secs():
return
self.tune_min_data_in_leaf()
if time_budget is not None and time_budget < t.elapsed_secs():
return
def run(self) -> None:
"""Perform the hyperparameter-tuning with given parameters."""
# Surpress log messages.
if self.auto_options["verbosity"] == 0:
optuna.logging.disable_default_handler()
self.lgbm_params["verbose"] = -1
self.lgbm_params["seed"] = 111
self.lgbm_kwargs["verbose_eval"] = False
# Handling aliases.
_handling_alias_parameters(self.lgbm_params)
# Sampling.
self.sample_train_set()
self.tune_feature_fraction()
self.tune_num_leaves()
self.tune_bagging()
self.tune_feature_fraction_stage2()
self.tune_regularization_factors()
self.tune_min_data_in_leaf()
