| 370 | # generate dataset pipline |
| 371 | def build_model_input(filename, batch_size, num_epochs): |
| 372 | def parse_csv(value): |
| 373 | tf.logging.info('Parsing {}'.format(filename)) |
| 374 | string_defaults = [[' '] for i in range(1, 19)] |
| 375 | label_defaults = [[0], [0]] |
| 376 | column_headers = INPUT_COLUMN |
| 377 | record_defaults = label_defaults + string_defaults |
| 378 | columns = tf.io.decode_csv(value, record_defaults=record_defaults) |
| 379 | all_columns = collections.OrderedDict(zip(column_headers, columns)) |
| 380 | labels = all_columns.pop(LABEL_COLUMN[0]) |
| 381 | all_columns.pop(BUY_COLUMN[0]) |
| 382 | features = all_columns |
| 383 | return features, labels |
| 384 | |
| 385 | def parse_parquet(value): |
| 386 | tf.logging.info('Parsing {}'.format(filename)) |