| 266 | # generate dataset pipline |
| 267 | def build_model_input(filename, batch_size, num_epochs): |
| 268 | def parse_csv(value): |
| 269 | tf.logging.info('Parsing {}'.format(filename)) |
| 270 | string_defaults = [[' '] for i in range(1, 19)] |
| 271 | label_defaults = [[0], [0]] |
| 272 | column_headers = INPUT_COLUMN |
| 273 | record_defaults = label_defaults + string_defaults |
| 274 | columns = tf.io.decode_csv(value, record_defaults=record_defaults) |
| 275 | all_columns = collections.OrderedDict(zip(column_headers, columns)) |
| 276 | labels = all_columns.pop(LABEL_COLUMN[0]) |
| 277 | all_columns.pop(BUY_COLUMN[0]) |
| 278 | features = all_columns |
| 279 | return features, labels |
| 280 | |
| 281 | def parse_parquet(value): |
| 282 | tf.logging.info('Parsing {}'.format(filename)) |