|
| string | __version__ = '1.0' |
| |
| string | __author__ = 'Saul Alonso-Monsalve' |
| |
| string | __email__ = "saul.alonso.monsalve@cern.ch" |
| |
| | stream |
| |
| | stdout |
| |
| | level |
| |
| | config = configparser.ConfigParser() |
| |
| | SEED = int(config['random']['seed']) |
| |
| | SHUFFLE = ast.literal_eval(config['random']['shuffle']) |
| |
| | IMAGES_PATH = config['images']['path'] |
| |
| | VIEWS = int(config['images']['views']) |
| |
| | PLANES = int(config['images']['planes']) |
| |
| | CELLS = int(config['images']['cells']) |
| |
| | STANDARDIZE = ast.literal_eval(config['images']['standardize']) |
| |
| | DATASET_PATH = config['dataset']['path'] |
| |
| | PARTITION_PREFIX = config['dataset']['partition_prefix'] |
| |
| | LABELS_PREFIX = config['dataset']['labels_prefix'] |
| |
| | CHECKPOINT_PATH = config['model']['checkpoint_path'] |
| |
| | CHECKPOINT_PREFIX = config['model']['checkpoint_prefix'] |
| |
| | CHECKPOINT_SAVE_MANY = ast.literal_eval(config['model']['checkpoint_save_many']) |
| |
| | CHECKPOINT_SAVE_BEST_ONLY = ast.literal_eval(config['model']['checkpoint_save_best_only']) |
| |
| | PRINT_SUMMARY = ast.literal_eval(config['model']['print_summary']) |
| |
| | BRANCHES = ast.literal_eval(config['model']['branches']) |
| |
| | PARALLELIZE = ast.literal_eval(config['model']['parallelize']) |
| |
| | OUTPUTS = int(config['model']['outputs']) |
| |
| | OUTPUT_PATH = config['test']['output_path'] |
| |
| | OUTPUT_PREFIX = config['test']['output_prefix'] |
| |
| | CUT_NUE = float(config['test']['cut_nue']) |
| |
| | CUT_NUMU = float(config['test']['cut_numu']) |
| |
| | CUT_NUTAU = float(config['test']['cut_nutau']) |
| |
| | CUT_NC = float(config['test']['cut_nc']) |
| |
| | TEST_BATCH_SIZE = int(config['test']['batch_size']) |
| |
| list | test_values = [] |
| |
| dictionary | TEST_PARAMS |
| |
| | partition = pickle.load(partition_file) |
| |
| | labels = pickle.load(labels_file) |
| |
| | prediction_generator = DataGenerator(**TEST_PARAMS).generate(labels, partition['test'], False) |
| |
| list | files = [f for f in os.listdir(CHECKPOINT_PATH) if os.path.isfile(os.path.join(CHECKPOINT_PATH, f))] |
| |
| | reverse |
| |
| | r = re.compile(CHECKPOINT_PREFIX[1:] + '-.*-.*.h5') |
| |
| | model |
| |
| list | is_antineutrino_target_names = ['neutrino', 'antineutrino', 'NULL'] |
| |
| list | flavour_target_names = ['CC Numu', 'CC Nue', 'CC Nutau', 'NC'] |
| |
| list | interaction_target_names = ['CC QE', 'CC Res', 'CC DIS', 'CC Other', 'NULL'] |
| |
| list | categories_target_names |
| |
| list | protons_target_names = ['0', '1', '2', '>2'] |
| |
| list | pions_target_names = ['0', '1', '2', '>2'] |
| |
| list | pizeros_target_names = ['0', '1', '2', '>2'] |
| |
| list | neutrons_target_names = ['0', '1', '2', '>2'] |
| |
| | Y_pred |
| |
| | y_pred_categories = np.argmax(Y_pred, axis=1).reshape((Y_pred.shape[0], 1)) |
| |
| | Y_pred_flavour = np.zeros((Y_pred.shape[0], 4)) |
| |
| | y_test_categories = np.array([12 if aux['y_value'] == 13 else aux['y_value'] for aux in test_values]).reshape(y_pred_categories.shape) |
| |
| | y_test_flavour = np.zeros(y_test_categories.shape, dtype=int) |
| |
| | p = Y_pred[i] |
| |
| | y_pred_flavour = np.argmax(Y_pred_flavour, axis=1) |
| |
| | flavour_conf_matrix = confusion_matrix(y_pred_flavour, y_test_flavour) |
| |
| | categories_conf_matrix = confusion_matrix(y_pred_categories, y_test_categories) |
| |
| | weighted_conf_matrix = np.zeros((4,4), dtype='float32') |
| |
| | cut_weighted_conf_matrix = np.zeros((4,4), dtype='float32') |
| |
| | pred_flavour = int(y_pred_flavour[sample]) |
| |
| | test_flavour = int(y_test_flavour[sample]) |
| |
| | weight = test_values[sample]['fEventWeight'] |
| |
| string | float_formatter = lambdax:"%.4f" |
| |
| | formatter |
| |
| | purity_conf_matrix = np.copy(cut_weighted_conf_matrix) |
| |
| | row_sum = np.sum(purity_conf_matrix[i]) |
| |
| | efficiency_conf_matrix = cut_weighted_conf_matrix.astype('float32')/np.add.reduce(weighted_conf_matrix) |
| |
| | y_pred_is_antineutrino = np.around(Y_pred[0]).reshape((Y_pred[0].shape[0], 1)).astype(int) |
| |
| | y_pred_interaction = np.argmax(Y_pred[2], axis=1).reshape((Y_pred[2].shape[0], 1)) |
| |
| | y_pred_protons = np.argmax(Y_pred[3], axis=1).reshape((Y_pred[3].shape[0], 1)) |
| |
| | y_pred_pions = np.argmax(Y_pred[4], axis=1).reshape((Y_pred[4].shape[0], 1)) |
| |
| | y_pred_pizeros = np.argmax(Y_pred[5], axis=1).reshape((Y_pred[5].shape[0], 1)) |
| |
| | y_pred_neutrons = np.argmax(Y_pred[6], axis=1).reshape((Y_pred[6].shape[0], 1)) |
| |
| | y_test_is_antineutrino = np.array([aux['y_value'][0] for aux in test_values]).reshape(y_pred_is_antineutrino.shape) |
| |
| | y_test_interaction = np.array([aux['y_value'][2] for aux in test_values]).reshape(y_pred_interaction.shape) |
| |
| | y_test_protons = np.array([aux['y_value'][3] for aux in test_values]).reshape(y_pred_protons.shape) |
| |
| | y_test_pions = np.array([aux['y_value'][4] for aux in test_values]).reshape(y_pred_pions.shape) |
| |
| | y_test_pizeros = np.array([aux['y_value'][5] for aux in test_values]).reshape(y_pred_pizeros.shape) |
| |
| | y_test_neutrons = np.array([aux['y_value'][6] for aux in test_values]).reshape(y_pred_neutrons.shape) |
| |
| | is_antineutrino_conf_matrix = confusion_matrix(y_pred_is_antineutrino, y_test_is_antineutrino) |
| |
| | interaction_conf_matrix = confusion_matrix(y_pred_interaction, y_test_interaction) |
| |
| | protons_conf_matrix = confusion_matrix(y_pred_protons, y_test_protons) |
| |
| | pions_conf_matrix = confusion_matrix(y_pred_pions, y_test_pions) |
| |
| | pizeros_conf_matrix = confusion_matrix(y_pred_pizeros, y_test_pizeros) |
| |
| | neutrons_conf_matrix = confusion_matrix(y_pred_neutrons, y_test_neutrons) |
| |
| dictionary | test_info |
| |
1.8.11
