def get_all_params():
all_params = []
for param in step2.get_param_list():
all_params.append(param)
for param in step3.get_param_list():
all_params.append(param)
#
_modelManager = ModelManager()
for param in utils.dict_get_param_list(_modelManager.get_properties()):
all_params.append(param)
#
for metric in utils.dict_get_param_list(
learn_evaluate_results.learning_metrics_template):
all_params.append(metric)
#
for base_params in utils.dict_get_param_list(
learn_evaluate_results.post_learning_metrics_template):
for postfix in ('val', 'test1', 'test2'):
current = base_params
current += '_'
current += postfix
all_params.append(current)
#
for obsolete_metrics in utils.dict_get_param_list(
learn_evaluate_results.obsolete_metrics_for_backward_compatibility
):
all_params.append(obsolete_metrics)
#
return all_params
def execute(dataset_name, dir_npy):
#
from train_data_generator import FCTrainDataGenerator
fcg = FCTrainDataGenerator()
fcg.load_compute_raw_data_additional_params(dataset_name)
fcg.load_compute_raw_data()
#
step2.step2_params['step2_target_class_col_name'] = 'target_class'
step2.step2_params['step2_profondeur_analyse'] = 3
step2.step2_params['step2_target_period'] = 'M15'
# paramètres spécifiques à 'generate_big_define_target'
step2.step2_params['step2_symbol_for_target'] = 'UsaInd'
step2.step2_params['step2_targets_classes_count'] = 3
step2.step2_params['step2_symbol_spread'] = 2.5
# step2_params['step2_targetLongShort'] = 20.0
# step2_params['step2_ratio_coupure'] = 1.3
# step2_params['step2_use_ATR'] = False
step2.step2_params['step2_targetLongShort'] = 0.95
step2.step2_params['step2_ratio_coupure'] = 1.1
step2.step2_params['step2_use_ATR'] = True
#
fcg.compute_target_additional_params(step2.step2_params)
fcg.compute_target()
#
model_manager = ModelManager()
#
# step3 parameters : unchanged during loop
#
step3.step3_params['step3_column_names_to_scale'] = []
step3.step3_params['step3_column_names_not_to_scale'] = [
'UsaInd_M15_time_slot',
'UsaInd_M15_pRSI_3', 'UsaInd_M15_pRSI_5', 'UsaInd_M15_pRSI_8', 'UsaInd_M15_pRSI_13', 'UsaInd_M15_pRSI_21']
step3.step3_params['step3_tests_by_class'] = 66
step3.step3_params['step3_idx_start'] = 0 # step3_idx_start = int(random.random()*1000)
#
for step3_recouvrement in (8, 13, 21, 34, 55, 89, 144, 233): # (2, 3, 5, 8, 13, 21, 34, 55, 89, 144, 233):
for step3_samples_by_class in (330, 660): # (330, 660, 990, 1320, 1650, 1980):
#
# step3 parameters : modified by this loop
#
step3.step3_params['step3_recouvrement'] = step3_recouvrement # 1 / proportion recouvrement
step3.step3_params['step3_time_depth'] = step3_recouvrement
step3.step3_params['step3_samples_by_class'] = step3_samples_by_class
#
try:
fcg.compute_learning_data_GRU_LSTM_Conv1D_additional_params(step3.step3_params)
learning_data = fcg.compute_learning_data_GRU_LSTM_Conv1D()
except :
print("fcg.create_step3_data failed. STOP")
return
#
# Model and learning parameters : unchanged during loop
#
_mm_dict = model_manager.get_properties()
#
_mm_dict['model_architecture'] = 'Conv1D_Dense'
_mm_dict['conv1D_block1_MaxPooling1D_pool_size'] = 2
_mm_dict['config_GRU_LSTM_units'] = 128
_mm_dict['config_Dense_units'] = 96
_mm_dict['dropout_rate'] = 0.5
_mm_dict['optimizer_name'] = 'adam'
_mm_dict['optimizer_modif_learning_rate'] = 0.75
#
_mm_dict['fit_batch_size'] = 32
_mm_dict['fit_epochs_max'] = 500
_mm_dict['fit_earlystopping_patience'] = 100
#
model_manager.update_properties(_mm_dict)
#
for conv1D_block1_filters in (55, 89, 144, 233, 377, 610, 987):
for conv1D_block1_kernel_size in (2, 3, 5):
#
# Model and learning parameters : modified by this loop
#
_mm_dict = model_manager.get_properties()
#
_mm_dict['conv1D_block1_filters'] = conv1D_block1_filters
_mm_dict['conv1D_block1_kernel_size'] = conv1D_block1_kernel_size
#
model_manager.update_properties(_mm_dict)
#
learn(dataset_name, dir_npy, model_manager, learning_data)
