add validation option to TrainDataset

src/gluonts/dataset/common.py

@@ -73,11 +73,12 @@ class SourceContext(NamedTuple):
 class TrainDatasets(NamedTuple):
  """
  A dataset containing two subsets, one to be used for training purposes, and
- the other for testing purposes, as well as metadata.
+ the other for validation and testing purposes, as well as metadata.
  """
 
  metadata: MetaData
  train: Dataset
+ validation: Optional[Dataset] = None
  test: Optional[Dataset] = None
 
  def save(
@@ -114,6 +115,11 @@ def save(
  test.mkdir(parents=True)
  writer.write_to_folder(self.test, test)
 
+ if self.validation is not None:
+ validation = path / "validation"
+ validation.mkdir(parents=True)
+ writer.write_to_folder(self.validation, validation)
+
 
 def infer_file_type(path):
  suffix = "".join(path.suffixes)
@@ -427,6 +433,7 @@ def __call__(self, data: DataEntry) -> DataEntry:
 def load_datasets(
  metadata: Path,
  train: Path,
+ validation: Optional[Path],
  test: Optional[Path],
  one_dim_target: bool = True,
  cache: bool = False,
@@ -442,6 +449,8 @@ def load_datasets(
  Path to the training dataset files.
  test
  Path to the test dataset files.
+ validation
+ Path to the validation dataset files.
  one_dim_target
  Whether to load FileDatasets as univariate target time series.
  cache
@@ -467,4 +476,17 @@ def load_datasets(
  else None
  )
 
- return TrainDatasets(metadata=meta, train=train_ds, test=test_ds)
+ validation_ds = (
+ FileDataset(
+ path=validation,
+ freq=meta.freq,
+ one_dim_target=one_dim_target,
+ cache=cache,
+ )
+ if validation
+ else None
+ )
+
+ return TrainDatasets(
+ metadata=meta, train=train_ds, validation=validation_ds, test=test_ds
+ )