#### END OF CONFIGURATION OPTION ####
y_train_values = [item for sublist in y_train_values for item in sublist]
y_train_predicted_values = [item for sublist in y_train_predicted_values for item in sublist]
# remove the -1 values (the ones that are censored)
tmp = [i for i in zip(y_train_values, y_train_predicted_values) if int(i[0]) != -1]
y_train_values = [i[0] for i in tmp]
y_train_predicted_values = [i[1] for i in tmp]
y_test_values = [item for sublist in y_test_values for item in sublist]
y_test_predicted_values = [item for sublist in y_test_predicted_values for item in sublist]
current_train_res, current_test_res = calc_results_and_plot(y_train_values, y_train_predicted_values,
y_test_values,
y_test_predicted_values, algo_name='NeuralNetwork',
visualize=PLOT,
title=f'Epochs: {epochs}, Validation iterations: {number_iterations}',
show=False)
# print(current_train_res)
# print(current_test_res)
if RECORD:
GRID_SEARCH_DIRECTORY = 'GridSearch_my_loss_factor_layers_neurons'
if not os.path.exists(GRID_SEARCH_DIRECTORY):
os.mkdir(GRID_SEARCH_DIRECTORY)
if not os.path.exists(GRID_SEARCH_DIRECTORY+'/'+current_configuration_str):
os.mkdir(GRID_SEARCH_DIRECTORY+'/'+current_configuration_str)
np.save(GRID_SEARCH_DIRECTORY+'/'+current_configuration_str+'/y_train_values.npy', y_train_values)
np.save(GRID_SEARCH_DIRECTORY+'/'+current_configuration_str+'/y_train_predicted_values.npy', y_train_predicted_values)
np.save(GRID_SEARCH_DIRECTORY+'/'+current_configuration_str+'/y_test_values.npy', y_test_values)
np.save(GRID_SEARCH_DIRECTORY+'/'+current_configuration_str+'/y_test_predicted_values.npy', y_test_predicted_values)
y_train_predicted_values = y_train_dict[
'alpha=0.01|n_estimators=20|min_child_weight=20|reg_lambda=20|max_depth=5'][
'y_train_predicted_values']
y_train_values = y_train_dict[
'alpha=0.01|n_estimators=20|min_child_weight=20|reg_lambda=20|max_depth=5'][
'y_train_values']
y_test_values = y_test_dict[
'alpha=0.01|n_estimators=20|min_child_weight=20|reg_lambda=20|max_depth=5'][
'y_test_values']
y_test_predicted_values = y_test_dict[
'alpha=0.01|n_estimators=20|min_child_weight=20|reg_lambda=20|max_depth=5'][
'y_test_predicted_values']
calc_results_and_plot(
y_train_values,
y_train_predicted_values,
y_test_values,
y_test_predicted_values,
'NewAlg\n alpha=0.01|n_estimators=20|min_child_weight=20|reg_lambda=20|max_depth=5',
visualize=True,
title='NewAlg')
print('**************\n')
print(betas_values)
print(removed_rows)
# betas_as_x = []
# y_train_rho = []
# y_test_rho = []
# for beta, value in betas_values.items():
# betas_as_x.append(beta)
# train = value[0]
# test = value[1]
# y_train_rho.append(
def calc_full_rho(configuration_stats,
epoch_to_use=None,
verbose_per_cv=False):
total_test_values = []
total_test_predicted_values = []
total_train_values = []
total_train_predicted_values = []
total_train_rho = []
total_train_mse = []
totral_train_pearson_rho = []
total_test_rho = []
total_test_mse = []
total_test_pearson_rho = []
for idx, (test_iter_cv_iter,
value) in enumerate(configuration_stats.items()):
print(test_iter_cv_iter)
test_real, test_predicted, train_real, train_predicted = get_data_from_last_epoch(
test_iter_cv_iter, value['epoch_list'], -10)
total_test_values += test_real.tolist()
total_test_predicted_values += test_predicted.tolist()
total_train_values += train_real.tolist()
total_train_predicted_values += train_predicted.tolist()
current_train, current_test = calc_results_and_plot(
train_real,
train_predicted,
test_real,
test_predicted,
algo_name='XGBoost',
visualize=False,
title='',
show=False)
total_train_rho.append(value['train_rho']['rho']['mean'])
total_train_mse.append(value['train_mse']['mse']['mean'])
totral_train_pearson_rho.append(
value['train_pearson_rho']['rho']['mean'])
total_test_rho.append(value['test_rho']['rho']['mean'])
total_test_mse.append(value['test_mse']['mse']['mean'])
total_test_pearson_rho.append(value['test_pearson_rho']['rho']['mean'])
if verbose_per_cv:
print(f'\tTrain\n\t\t{current_train}')
if (idx + 1) % len(configuration_stats) == 0:
current_train_res, current_test_res = calc_results_and_plot(
total_train_values,
total_train_predicted_values,
total_test_values,
total_test_predicted_values,
algo_name='XGBoost',
visualize=False,
title='',
show=False)
print(f'Total RHO')
print(f'\tTrain\n\t\t{current_train_res}')
print(f'\tTest\n\t\t{current_test_res}')
total_test_values = []
total_test_predicted_values = []
total_train_values = []
total_train_predicted_values = []
if verbose_per_cv:
print(f'\n\tTrain\n\t\t')
print(f'rho: {get_stats_for_array(np.array(total_train_rho))}')
print(f'mse: {get_stats_for_array(np.array(total_train_mse))}')
print(
f'pearson: {get_stats_for_array(np.array(totral_train_pearson_rho))}'
)
print(f'\tTest\n\t\t')
print(f'rho: {get_stats_for_array(np.array(total_test_rho))}')
print(f'mse: {get_stats_for_array(np.array(total_test_mse))}')
print(
f'pearson: {get_stats_for_array(np.array(total_test_pearson_rho))}'
)
print(f'\n')