filter

Module defining the filter object.

Classes:

Name	Description
Filter	Filter class, used to filter a dataset for causal explanation on a subset.

Filter(name: str, fitted_parquet_dataset: FittedParquetDataset, criteria: list[NumericalCriterion | CategoricalCriterion | MultivariateCriterion | TimeseriesBoxCriterion] | None = None, row_indexes: list[int] | None = None)

Filter class, used to filter a dataset for causal explanation on a subset.

Initialize the filter object.

Parameters:

Name	Type	Description	Default
name	str	The filter name.	required
fitted_parquet_dataset	FittedParquetDataset	The dataset to filter will be applied to.	required
criteria	list[NumericalCriterion | CategoricalCriterion] | None	An optional list of filter criteria to filter by feature.	None
row_indexes	list[int] | None	An optional list of row indexes to filter by sample.	None.

Methods:

Name	Description
add_criterion	Add criterion.
add_criteria	Add many criteria.
apply	Generate/update a remote filter.
__len__	Get filter's result size.

Attributes:

Name	Type	Description
name
criteria
row_indexes
dataset
id	str | None	Get id.

Source code in src/xpdeep/filtering/filter.py

def __init__(
    self,
    name: str,
    fitted_parquet_dataset: FittedParquetDataset,
    criteria: list[NumericalCriterion | CategoricalCriterion | MultivariateCriterion | TimeseriesBoxCriterion]
    | None = None,
    row_indexes: list[int] | None = None,
) -> None:
    """
    Initialize the filter object.

    Parameters
    ----------
    name : str
        The filter name.
    fitted_parquet_dataset : FittedParquetDataset
        The dataset to filter will be applied to.
    criteria : list[NumericalCriterion | CategoricalCriterion] | None
        An optional list of filter criteria to filter by feature.
    row_indexes : list[int] | None, default None.
        An optional list of row indexes to filter by sample.
    """
    self._id: str | None = None
    self.name = name
    self.criteria = criteria if criteria is not None else []
    self.row_indexes = row_indexes
    self.dataset = fitted_parquet_dataset

name = name

criteria = criteria if criteria is not None else []

row_indexes = row_indexes

dataset = fitted_parquet_dataset

id: str | None

Get id.

add_criterion(criterion: NumericalCriterion | CategoricalCriterion) -> None

Add criterion.

Source code in src/xpdeep/filtering/filter.py

def add_criterion(self, criterion: NumericalCriterion | CategoricalCriterion) -> None:
    """Add criterion."""
    self.criteria.append(criterion)

add_criteria(*args: NumericalCriterion | CategoricalCriterion) -> None

Add many criteria.

Source code in src/xpdeep/filtering/filter.py

def add_criteria(self, *args: NumericalCriterion | CategoricalCriterion) -> None:
    """Add many criteria."""
    self.criteria += list(args)

apply() -> None

Generate/update a remote filter.

Source code in src/xpdeep/filtering/filter.py

def apply(self) -> None:
    """Generate/update a remote filter."""
    if self._id is None:
        self._create()
    else:
        self._update()

__len__() -> int

Get filter's result size.

Source code in src/xpdeep/filtering/filter.py

@initialized_client_verification
@initialized_project_verification
def __len__(self) -> int:
    """Get filter's result size."""
    if self.id is None:
        msg = "This filer is not applied yet. Call `apply` method before computing filter's results size."
        raise ApiError(msg)

    return cast(
        FilterSizeResponseBody,
        get_filter_size.sync(
            project_id=Project.CURRENT.get().model.id, id=self.id, client=ClientFactory.CURRENT.get()()
        ),
    ).size