DataSplitter

class openstef_models.utils.data_split.DataSplitter(**data: Any) → None

Bases: BaseConfig

Handles splitting of time series data into train, validation, and test sets.

Supports stratified splitting to ensure representative data distribution across splits, particularly for extreme values in forecasting scenarios.

Parameters:

data (Any)

val_fraction: float

test_fraction: float

stratification_fraction: float

min_days_for_stratification: int

random_state: int

split_dataset(data: T, data_val: T | None = None, data_test: T | None = None, target_column: str = 'load') → tuple[T, T | None, T | None]

Prepare and split input data into train, validation, and test sets.

Parameters:

• data (TypeVar(T, bound= TimeSeriesDataset)) – Full dataset to split.

• data_val (Optional[TypeVar(T, bound= TimeSeriesDataset)]) – Optional pre-split validation data.

• data_test (Optional[TypeVar(T, bound= TimeSeriesDataset)]) – Optional pre-split test data.

• target_column (str) – Column name containing the target variable for stratification.

• data

• data_val

• data_test

• target_column

Returns:

Tuple of (train_data, val_data, test_data) where val_data and test_data may be None.

Return type:

tuple[TypeVar(T, bound= TimeSeriesDataset), Optional[TypeVar(T, bound= TimeSeriesDataset)], Optional[TypeVar(T, bound= TimeSeriesDataset)]]

model_config: ClassVar[ConfigDict] = {'arbitrary_types_allowed': False, 'extra': 'ignore', 'protected_namespaces': ()}

Configuration for the model, should be a dictionary conforming to [ConfigDict][pydantic.config.ConfigDict].