feature

The features module provides a list of feature object to be used to define the dataset structure.

Modules:

Name	Description
abstracts	Feature abstract classes module.
augmentation	Data augmentation package.
feature	Represent a feature object.
feature_types	Access all feature types.

Classes:

Name	Description
BaseFeature	Base Feature used if no convenient available feature.
ExplainableFeature	Define a feature object.
IndexMetadata	Represents an index metadata, automatically added to the schema, cf doc dataset/concept.

BaseFeature

Base Feature used if no convenient available feature.

Parameters:

Name	Type	Description	Default
name	str		required
id	str | None		None
is_target	bool		False

ExplainableFeature

Define a feature object.

Parameters:

Name	Type	Description	Default
id	UUID | None	Feature Identifier	None
is_target	bool	Whether the feature is target or not.	False
feature_type	FeatureTypeUnion		required
UnivariateTimeSeriesFeature		Feature type.	required
preprocessor	TorchPreprocessor | SklearnPreprocessor | IdentityPreprocessor	The feature preprocessor function used to preprocess the data.	required
name	str		required
augmentation	FeatureAugmentation | None		None

Methods:

Name	Description
to_model	Convert to FeatureInsert instance.
stable_hash	Return a hash, without the id.

Attributes:

Name	Type	Description
feature_type	FeatureTypeUnion
preprocessor	TorchPreprocessor | SklearnPreprocessor | IdentityPreprocessor | BoundingBoxesPreprocessor
augmentation	FeatureAugmentation | None

feature_type: FeatureTypeUnion = field(kw_only=True)

preprocessor: TorchPreprocessor | SklearnPreprocessor | IdentityPreprocessor | BoundingBoxesPreprocessor = field(kw_only=True)

augmentation: FeatureAugmentation | None = field(kw_only=True, default=None)

to_model() -> FeatureInsert

Convert to FeatureInsert instance.

Source code in src/xpdeep/dataset/feature/feature.py

def to_model(self) -> FeatureInsert:
    """Convert to FeatureInsert instance."""
    return FeatureInsert(
        type_="FEATURE",
        name=self.name,
        feature_type=self.feature_type.to_model(),
        is_target=self.is_target,
        preprocessor=self.preprocessor.to_model(),
        augmentation=self.augmentation.to_model() if self.augmentation is not None else None,
    )

stable_hash() -> str

Return a hash, without the id.

Source code in src/xpdeep/dataset/feature/feature.py

def stable_hash(self) -> str:
    """Return a hash, without the id."""
    augmentation_hash = self.augmentation.stable_hash() if self.augmentation is not None else None
    return str(
        sha256(
            f"{self.feature_type.stable_hash()}_{self.preprocessor.stable_hash()}_{augmentation_hash}_{self.is_target}".encode()
        ).hexdigest()
    )

IndexMetadata

Represents an index metadata, automatically added to the schema, cf doc dataset/concept.

Parameters:

Name	Type	Description	Default
name	str		required
id	str | None		None
feature_type	str		'xpdeep_index'

Methods:

Name	Description
to_model	Convert to IndexMetadataInsert instance.
stable_hash	Return a hash, without the id.

Attributes:

Name	Type	Description
name	str
id	str | None
feature_type	str

name: str

id: str | None = None

feature_type: str = 'xpdeep_index'

to_model() -> IndexMetadataInsert

Convert to IndexMetadataInsert instance.

Source code in src/xpdeep/dataset/feature/feature.py

def to_model(self) -> IndexMetadataInsert:
    """Convert to IndexMetadataInsert instance."""
    return IndexMetadataInsert(name=self.name, type_="INDEX_METADATA")

stable_hash() -> str

Return a hash, without the id.

Source code in src/xpdeep/dataset/feature/feature.py

def stable_hash(self) -> str:
    """Return a hash, without the id."""
    return str(sha256(f"{self.name}_{self.feature_type}".encode()).hexdigest())