Features

Registering an Entity

Class decorator for registering entities and their associated features and labels.

Examples

@ff.entity
class User:
    avg_transactions = ff.Feature()
    fraudulent = ff.Label()

Returns:

Name	Type	Description
`entity`	`class`	Decorated class

Feature

Feature registration object.

Example

@ff.entity
class Customer:
# Register a column from a transformation as a feature
    transaction_amount = ff.Feature(
        fare_per_family_member[["CustomerID", "Amount", "Transaction Time"]],
        variant="quickstart",
        type=ff.Float64,
        inference_store=redis,
    )

Parameters:

Name	Type	Description	Default
`transformation_args`	`tuple`	A transformation or source function and the columns name in the format: [[, , ]].	required
`variant`	`str`	An optional variant name for the feature.	`''`
`type`	`Union[ScalarType, str]`	The type of the value in for the feature.	required
`inference_store`	`Union[str, OnlineProvider, FileStoreProvider]`	Where to store for online serving.	`''`

MultiFeature

Registering multiple features from the same table. The name of each feature is the name of the column in the table.

Example

# Register a file or table from an offline provider as a dataset

client = ff.Client()
df = client.dataframe(dataset)

@ff.entity
class Customer:
# Register multiple columns from a dataset as features
    transaction_features = ff.MultiFeature(
        dataset,
        df,
        variant="quickstart",
        inference_store=redis,
        entity_column="CustomerID",
        timestamp_column="Timestamp",
        exclude_columns=["TransactionID", "IsFraud"],
        inference_store=redis,
    )

Parameters:

Name	Type	Description	Default
`dataset`	`SourceVariant`	The dataset to register features from	required
`df`	`DataFrame`	The client.dataframe to register features from	required
`include_columns`	`List[str]`	List of columns to be registered as features	`None`
`exclude_columns`	`List[str]`	List of columns to be excluded from registration	`None`
`entity_column`	`Union[Entity, str]`	The name of the column in the source to be used as the entity	required
`variant`	`str`	An optional variant name for the feature.	`''`
`inference_store`	`Union[str, OnlineProvider, FileStoreProvider]`	Where to store for online serving.	`''`

Label

Label registration object.

Example

@ff.entity
class Customer:
# Register a column from a transformation as a label
    transaction_amount = ff.Label(
        fare_per_family_member[["CustomerID", "Amount", "Transaction Time"]],
        variant="quickstart",
        type=ff.Float64
    )

Parameters:

Name	Type	Description	Default
`transformation_args`	`tuple`	A transformation or source function and the columns name in the format: [[, , ]]	required
`variant`	`str`	An optional variant name for the label.	`''`
`type`	`Union[ScalarType, str]`	The type of the value in for the label.	required

Embedding

Embedding Feature registration object.

Example

@ff.entity
class Speaker:
# Register a column from a transformation as a label
    transaction_amount = ff.Embedding(
        vectorize_comments[["PK", "Vector"]],
        dims=384,
        vector_db=pinecone,
        description="Embeddings created from speakers' comments in episodes",
        variant="v1"
    )

Parameters:

Name	Type	Description	Default
`transformation_args`	`tuple`	A transformation or source function and the columns name in the format: [[, ]]	required
`dims`	`int`	Dimensionality of the embedding.	required
`vector_db`	`Union[str, OnlineProvider]`	The name of the vector database to store the embeddings in.	required
`variant`	`str`	An optional variant name for the feature.	`''`
`description`	`str`	An optional description for the feature.	`''`