Feature Stores

zenml.feature_stores

A feature store enables an offline and online serving of feature data.

Feature stores allow data teams to serve data via an offline store and an online low-latency store where data is kept in sync between the two. It also offers a centralized registry where features (and feature schemas) are stored for use within a team or wider organization.

As a data scientist working on training your model, your requirements for how you access your batch / 'offline' data will almost certainly be different from how you access that data as part of a real-time or online inference setting. Feast solves the problem of developing train-serve skew where those two sources of data diverge from each other.

Attributes

__all__ = ['BaseFeatureStore'] module-attribute

Classes

BaseFeatureStore(name: str, id: UUID, config: StackComponentConfig, flavor: str, type: StackComponentType, user: Optional[UUID], created: datetime, updated: datetime, labels: Optional[Dict[str, Any]] = None, connector_requirements: Optional[ServiceConnectorRequirements] = None, connector: Optional[UUID] = None, connector_resource_id: Optional[str] = None, *args: Any, **kwargs: Any)

Bases: StackComponent, ABC

Base class for all ZenML feature stores.

Source code in src/zenml/stack/stack_component.py

def __init__(
    self,
    name: str,
    id: UUID,
    config: StackComponentConfig,
    flavor: str,
    type: StackComponentType,
    user: Optional[UUID],
    created: datetime,
    updated: datetime,
    labels: Optional[Dict[str, Any]] = None,
    connector_requirements: Optional[ServiceConnectorRequirements] = None,
    connector: Optional[UUID] = None,
    connector_resource_id: Optional[str] = None,
    *args: Any,
    **kwargs: Any,
):
    """Initializes a StackComponent.

    Args:
        name: The name of the component.
        id: The unique ID of the component.
        config: The config of the component.
        flavor: The flavor of the component.
        type: The type of the component.
        user: The ID of the user who created the component.
        created: The creation time of the component.
        updated: The last update time of the component.
        labels: The labels of the component.
        connector_requirements: The requirements for the connector.
        connector: The ID of a connector linked to the component.
        connector_resource_id: The custom resource ID to access through
            the connector.
        *args: Additional positional arguments.
        **kwargs: Additional keyword arguments.

    Raises:
        ValueError: If a secret reference is passed as name.
    """
    if secret_utils.is_secret_reference(name):
        raise ValueError(
            "Passing the `name` attribute of a stack component as a "
            "secret reference is not allowed."
        )

    self.id = id
    self.name = name
    self._config = config
    self.flavor = flavor
    self.type = type
    self.user = user
    self.created = created
    self.updated = updated
    self.labels = labels
    self.connector_requirements = connector_requirements
    self.connector = connector
    self.connector_resource_id = connector_resource_id
    self._connector_instance: Optional[ServiceConnector] = None

Attributes

config: BaseFeatureStoreConfig property

Returns the BaseFeatureStoreConfig config.

Returns:

Type	Description
BaseFeatureStoreConfig	The configuration.

Functions

get_historical_features(entity_df: Any, features: List[str], full_feature_names: bool = False) -> Any abstractmethod

Returns the historical features for training or batch scoring.

Parameters:

Name	Type	Description	Default
entity_df	Any	The entity DataFrame or entity name.	required
features	List[str]	The features to retrieve.	required
full_feature_names	bool	Whether to return the full feature names.	False

Returns:

Type	Description
Any	The historical features.

Source code in src/zenml/feature_stores/base_feature_store.py

@abstractmethod
def get_historical_features(
    self,
    entity_df: Any,
    features: List[str],
    full_feature_names: bool = False,
) -> Any:
    """Returns the historical features for training or batch scoring.

    Args:
        entity_df: The entity DataFrame or entity name.
        features: The features to retrieve.
        full_feature_names: Whether to return the full feature names.

    Returns:
        The historical features.
    """

get_online_features(entity_rows: List[Dict[str, Any]], features: List[str], full_feature_names: bool = False) -> Dict[str, Any] abstractmethod

Returns the latest online feature data.

Parameters:

Name	Type	Description	Default
entity_rows	List[Dict[str, Any]]	The entity rows to retrieve.	required
features	List[str]	The features to retrieve.	required
full_feature_names	bool	Whether to return the full feature names.	False

Returns:

Type	Description
Dict[str, Any]	The latest online feature data as a dictionary.

Source code in src/zenml/feature_stores/base_feature_store.py

@abstractmethod
def get_online_features(
    self,
    entity_rows: List[Dict[str, Any]],
    features: List[str],
    full_feature_names: bool = False,
) -> Dict[str, Any]:
    """Returns the latest online feature data.

    Args:
        entity_rows: The entity rows to retrieve.
        features: The features to retrieve.
        full_feature_names: Whether to return the full feature names.

    Returns:
        The latest online feature data as a dictionary.
    """

Modules

base_feature_store

The base class for feature stores.

Classes

BaseFeatureStore(name: str, id: UUID, config: StackComponentConfig, flavor: str, type: StackComponentType, user: Optional[UUID], created: datetime, updated: datetime, labels: Optional[Dict[str, Any]] = None, connector_requirements: Optional[ServiceConnectorRequirements] = None, connector: Optional[UUID] = None, connector_resource_id: Optional[str] = None, *args: Any, **kwargs: Any)

Bases: StackComponent, ABC

Base class for all ZenML feature stores.

Source code in src/zenml/stack/stack_component.py

def __init__(
    self,
    name: str,
    id: UUID,
    config: StackComponentConfig,
    flavor: str,
    type: StackComponentType,
    user: Optional[UUID],
    created: datetime,
    updated: datetime,
    labels: Optional[Dict[str, Any]] = None,
    connector_requirements: Optional[ServiceConnectorRequirements] = None,
    connector: Optional[UUID] = None,
    connector_resource_id: Optional[str] = None,
    *args: Any,
    **kwargs: Any,
):
    """Initializes a StackComponent.

    Args:
        name: The name of the component.
        id: The unique ID of the component.
        config: The config of the component.
        flavor: The flavor of the component.
        type: The type of the component.
        user: The ID of the user who created the component.
        created: The creation time of the component.
        updated: The last update time of the component.
        labels: The labels of the component.
        connector_requirements: The requirements for the connector.
        connector: The ID of a connector linked to the component.
        connector_resource_id: The custom resource ID to access through
            the connector.
        *args: Additional positional arguments.
        **kwargs: Additional keyword arguments.

    Raises:
        ValueError: If a secret reference is passed as name.
    """
    if secret_utils.is_secret_reference(name):
        raise ValueError(
            "Passing the `name` attribute of a stack component as a "
            "secret reference is not allowed."
        )

    self.id = id
    self.name = name
    self._config = config
    self.flavor = flavor
    self.type = type
    self.user = user
    self.created = created
    self.updated = updated
    self.labels = labels
    self.connector_requirements = connector_requirements
    self.connector = connector
    self.connector_resource_id = connector_resource_id
    self._connector_instance: Optional[ServiceConnector] = None

Attributes

config: BaseFeatureStoreConfig property

Returns the BaseFeatureStoreConfig config.

Returns:

Type	Description
BaseFeatureStoreConfig	The configuration.

Functions

get_historical_features(entity_df: Any, features: List[str], full_feature_names: bool = False) -> Any abstractmethod

Returns the historical features for training or batch scoring.

Parameters:

Name	Type	Description	Default
entity_df	Any	The entity DataFrame or entity name.	required
features	List[str]	The features to retrieve.	required
full_feature_names	bool	Whether to return the full feature names.	False

Returns:

Type	Description
Any	The historical features.

Source code in src/zenml/feature_stores/base_feature_store.py

@abstractmethod
def get_historical_features(
    self,
    entity_df: Any,
    features: List[str],
    full_feature_names: bool = False,
) -> Any:
    """Returns the historical features for training or batch scoring.

    Args:
        entity_df: The entity DataFrame or entity name.
        features: The features to retrieve.
        full_feature_names: Whether to return the full feature names.

    Returns:
        The historical features.
    """

get_online_features(entity_rows: List[Dict[str, Any]], features: List[str], full_feature_names: bool = False) -> Dict[str, Any] abstractmethod

Returns the latest online feature data.

Parameters:

Name	Type	Description	Default
entity_rows	List[Dict[str, Any]]	The entity rows to retrieve.	required
features	List[str]	The features to retrieve.	required
full_feature_names	bool	Whether to return the full feature names.	False

Returns:

Type	Description
Dict[str, Any]	The latest online feature data as a dictionary.

Source code in src/zenml/feature_stores/base_feature_store.py

@abstractmethod
def get_online_features(
    self,
    entity_rows: List[Dict[str, Any]],
    features: List[str],
    full_feature_names: bool = False,
) -> Dict[str, Any]:
    """Returns the latest online feature data.

    Args:
        entity_rows: The entity rows to retrieve.
        features: The features to retrieve.
        full_feature_names: Whether to return the full feature names.

    Returns:
        The latest online feature data as a dictionary.
    """

BaseFeatureStoreConfig(warn_about_plain_text_secrets: bool = False, **kwargs: Any)

Bases: StackComponentConfig

Base config for feature stores.

Source code in src/zenml/stack/stack_component.py

def __init__(
    self, warn_about_plain_text_secrets: bool = False, **kwargs: Any
) -> None:
    """Ensures that secret references don't clash with pydantic validation.

    StackComponents allow the specification of all their string attributes
    using secret references of the form `{{secret_name.key}}`. This however
    is only possible when the stack component does not perform any explicit
    validation of this attribute using pydantic validators. If this were
    the case, the validation would run on the secret reference and would
    fail or in the worst case, modify the secret reference and lead to
    unexpected behavior. This method ensures that no attributes that require
    custom pydantic validation are set as secret references.

    Args:
        warn_about_plain_text_secrets: If true, then warns about using
            plain-text secrets.
        **kwargs: Arguments to initialize this stack component.

    Raises:
        ValueError: If an attribute that requires custom pydantic validation
            is passed as a secret reference, or if the `name` attribute
            was passed as a secret reference.
    """
    for key, value in kwargs.items():
        try:
            field = self.__class__.model_fields[key]
        except KeyError:
            # Value for a private attribute or non-existing field, this
            # will fail during the upcoming pydantic validation
            continue

        if value is None:
            continue

        if not secret_utils.is_secret_reference(value):
            if (
                secret_utils.is_secret_field(field)
                and warn_about_plain_text_secrets
            ):
                logger.warning(
                    "You specified a plain-text value for the sensitive "
                    f"attribute `{key}` for a `{self.__class__.__name__}` "
                    "stack component. This is currently only a warning, "
                    "but future versions of ZenML will require you to pass "
                    "in sensitive information as secrets. Check out the "
                    "documentation on how to configure your stack "
                    "components with secrets here: "
                    "https://docs.zenml.io/getting-started/deploying-zenml/secret-management"
                )
            continue

        if pydantic_utils.has_validators(
            pydantic_class=self.__class__, field_name=key
        ):
            raise ValueError(
                f"Passing the stack component attribute `{key}` as a "
                "secret reference is not allowed as additional validation "
                "is required for this attribute."
            )

    super().__init__(**kwargs)

BaseFeatureStoreFlavor

Bases: Flavor

Base class for all ZenML feature store flavors.

Attributes

config_class: Type[BaseFeatureStoreConfig] property

Config class for this flavor.

Returns:

Type	Description
Type[BaseFeatureStoreConfig]	The config class.

implementation_class: Type[BaseFeatureStore] abstractmethod property

Implementation class.

Returns:

Type	Description
Type[BaseFeatureStore]	The implementation class.

type: StackComponentType property

Returns the flavor type.

Returns:

Type	Description
StackComponentType	The flavor type.