SequenceDataset

class rectools.models.nn.transformers.data_preparator.SequenceDataset(sessions: List[List[int]], weights: List[List[float]], extras: Optional[Dict[str, List[Any]]] = None)[source]

Bases: Dataset

Dataset for sequential data.

Parameters

  • sessions (List[List[int]]) – User sessions in the form of sequences of items ids.

  • weights (List[List[float]]) – Weight of each interaction from the session.

  • extras (Optional[Dict[str, List[Any]]]) –

Inherited-members

Methods

from_interactions(interactions[, sort_users])

Group interactions by user.
classmethod from_interactions(interactions: DataFrame, sort_users: bool = False) → SequenceDataset[source]

Group interactions by user. Construct SequenceDataset from grouped interactions.

Parameters

  • interactions (pd.DataFrame) – User-item interactions.

  • sort_users (bool) –

Return type

SequenceDataset