calc_debiased_fit_task

rectools.metrics.debias.calc_debiased_fit_task(metrics: Iterable[DebiasableMetrikAtK], interactions: DataFrame, prev_debiased_interactions: Optional[Dict[DebiasConfig, DataFrame]] = None) → Dict[DebiasConfig, Tuple[int, DataFrame]][source]

Calculate for each of the unique debias configs k_max and debiased interactions to then apply them in the fit methods of the corresponding metrics.

Parameters

  • metrics (tp.Iteraple[DebiasableMetrikAtK]) – Dict of metric objects to calculate, where key is metric name and value is metric object.

  • interactions (pd.DataFrame) – Interactions or merging table with columns Columns.User, Columns.Item, Columns.Rank (for merging). Obligatory only for some types of metrics.

  • prev_debiased_interactions (dict(DebiasConfig->pd.DataFrame]), optinonal) – Debiased interactions for certain debias configs calculated earlier.

Returns

Dictionary, where key is debias config and values are a tuple of the corresponding k_max and debiased interactions.

Return type

dict(DebiasConfig->list[(int | pd.DataFrame)])