# 'random_normal_initializer_stdev': 0.00025797511482927632,
# 'rate_of_learning': 0.20172634121590136}
# persist the optimized configuration to a file
persist_results(
optimized_configuration,
optimized_config_directory + '/' + model_identifier + '.txt')
# optimized_configuration = read_optimal_hyperparameter_values(optimized_config_directory + '/' + model_identifier + '.txt')
# get the validation errors for the best hyperparameter configs
smape_error, smape_error_list = train_model(optimized_configuration)
# print(smape_error_list)
# write the final list of validation errors to a file
validation_errors_file = model_training_configs.VALIDATION_ERRORS_DIRECTORY + model_identifier + ".csv"
with open(validation_errors_file, "w") as output:
writer = csv.writer(output, lineterminator='\n')
writer.writerow(smape_error_list)
print("Optimized configuration: {}".format(optimized_configuration))
print("Optimized Value: {}\n".format(smape_error))
# test the model
for i in range(1, 11):
args.seed = i
testing(args, optimized_configuration)
# testing(args, optimized_configuration)
# select the optimizer
if optimizer == "cocob":
optimizer_fn = cocob_optimizer_fn
elif optimizer == "adagrad":
optimizer_fn = adagrad_optimizer_fn
elif optimizer == "adam":
optimizer_fn = adam_optimizer_fn
optimized_configuration = {
'num_hidden_layers':
optimized_params['num_hidden_layers'],
'cell_dimension':
optimized_params['cell_dimension'],
'l2_regularization':
optimized_params['l2_regularization'],
'gaussian_noise_stdev':
optimized_params['gaussian_noise_stdev'],
'random_normal_initializer_stdev':
optimized_params['random_normal_initializer_stdev'],
'minibatch_size':
optimized_params['minibatch_size'],
'max_epoch_size':
optimized_params['max_epoch_size'],
'max_num_epochs':
optimized_params['max_num_epochs'],
'learning_rate':
''
}
testing(args, optimized_configuration, "validation")
'seed': seed,
'cell_type': cell_type,
'without_stl_decomposition': without_stl_decomposition
}
model_trainer = StackingModelTrainer(**model_kwargs)
# read the initial hyperparamter configurations from the file
hyperparameter_values_dic = read_initial_hyperparameter_values(
initial_hyperparameter_values_file)
optimized_configuration = smac()
# persist the optimized configuration to a file
persist_results(
optimized_configuration,
optimized_config_directory + '/' + model_identifier + '.txt')
# get the validation errors for the best hyperparameter configs
smape_error, smape_error_list = train_model(optimized_configuration)
# write the final list of validation errors to a file
validation_errors_file = model_training_configs.VALIDATION_ERRORS_DIRECTORY + model_identifier + ".csv"
with open(validation_errors_file, "w") as output:
writer = csv.writer(output, lineterminator='\n')
writer.writerow(smape_error_list)
print("Optimized configuration: {}".format(optimized_configuration))
print("Optimized Value: {}\n".format(smape_error))
testing(args, optimized_configuration, "test")