Skip to content

Feature Store#

Retrieval#

[source]

get_feature_store#

Connection.get_feature_store(name=None)

Get a reference to a feature store to perform operations on.

Defaulting to the project's default feature store. Shared feature stores can be retrieved by passing the name argument.

Arguments

  • name str: The name of the feature store, defaults to None.

Returns

FeatureStore. A feature store handle object to perform operations on.

Properties#

[source]

description#

Description of the feature store.

[source]

hive_endpoint#

Hive endpoint for the offline feature store.

[source]

id#

Id of the feature store.

[source]

mysql_server_endpoint#

MySQL server endpoint for the online feature store.

[source]

name#

Name of the feature store.

[source]

offline_featurestore_name#

Name of the offline feature store database.

[source]

online_enabled#

Indicator whether online feature store is enabled.

[source]

online_featurestore_name#

Name of the online feature store database.

[source]

project_id#

Id of the project in which the feature store is located.

[source]

project_name#

Name of the project in which the feature store is located.

Methods#

[source]

create_feature_group#

FeatureStore.create_feature_group(
    name,
    version=None,
    description="",
    online_enabled=False,
    time_travel_format="HUDI",
    partition_key=[],
    primary_key=[],
    hudi_precombine_key=None,
    features=[],
    statistics_config=None,
)

Create a feature group metadata object.

Lazy

This method is lazy and does not persist any metadata or feature data in the feature store on its own. To persist the feature group and save feature data along the metadata in the feature store, call the save() method with a DataFrame.

Arguments

  • name str: Name of the feature group to create.
  • version Optional[int]: Version of the feature group to retrieve, defaults to None and will create the feature group with incremented version from the last version in the feature store.
  • description Optional[str]: A string describing the contents of the feature group to improve discoverability for Data Scientists, defaults to empty string "".
  • online_enabled Optional[bool]: Define whether the feature group should be made available also in the online feature store for low latency access, defaults to False.
  • time_travel_format Optional[str]: Format used for time travel, defaults to "HUDI".
  • partition_key Optional[List[str]]: A list of feature names to be used as partition key when writing the feature data to the offline storage, defaults to empty list [].
  • primary_key Optional[List[str]]: A list of feature names to be used as primary key for the feature group. This primary key can be a composite key of multiple features and will be used as joining key, if not specified otherwise. Defaults to empty list [], and the first column of the DataFrame will be used as primary key.
  • hudi_precombine_key Optional[str]: A feature name to be used as a precombine key for the "HUDI" feature group. Defaults to None. If feature group has time travel format "HUDI" and hudi precombine key was not specified then the first primary key of the feature group will be used as hudi precombine key.
  • features Optional[List[hsfs.feature.Feature]]: Optionally, define the schema of the feature group manually as a list of Feature objects. Defaults to empty list [] and will use the schema information of the DataFrame provided in the save method.
  • statistics_config Optional[Union[hsfs.StatisticsConfig, bool, dict]]: A configuration object, or a dictionary with keys "enabled" to generally enable descriptive statistics computation for this feature group, "correlations" to turn on feature correlation computation and "histograms" to compute feature value frequencies. The values should be booleans indicating the setting. To fully turn off statistics computation pass statistics_config=False. Defaults to None and will compute only descriptive statistics.

Returns

FeatureGroup. The feature group metadata object.

[source]

create_on_demand_feature_group#

FeatureStore.create_on_demand_feature_group(
    name,
    storage_connector,
    query=None,
    data_format=None,
    path="",
    options={},
    version=None,
    description="",
    features=[],
    statistics_config=None,
)

Create a on-demand feature group metadata object.

Lazy

This method is lazy and does not persist any metadata or feature data in the feature store on its own. To persist the feature group and save feature data along the metadata in the feature store, call the save() method.

Arguments

  • name str: Name of the on-demand feature group to create.
  • query Optional[str]: A string containing a SQL query valid for the target data source. the query will be used to pull data from the data sources when the feature group is used.
  • data_format Optional[str]: If the on-demand feature groups refers to a directory with data, the data format to use when reading it
  • path Optional[str]: The location within the scope of the storage connector, from where to read the data for the on-demand feature group
  • storage_connector hsfs.StorageConnector: the storage connector to use to establish connectivity with the data source.
  • version Optional[int]: Version of the on-demand feature group to retrieve, defaults to None and will create the feature group with incremented version from the last version in the feature store.
  • description Optional[str]: A string describing the contents of the on-demand feature group to improve discoverability for Data Scientists, defaults to empty string "".
  • features Optional[List[hsfs.feature.Feature]]: Optionally, define the schema of the on-demand feature group manually as a list of Feature objects. Defaults to empty list [] and will use the schema information of the DataFrame resulting by executing the provided query against the data source.
  • statistics_config Optional[Union[hsfs.StatisticsConfig, bool, dict]]: A configuration object, or a dictionary with keys "enabled" to generally enable descriptive statistics computation for this on-demand feature group, "correlations" to turn on feature correlation computation and "histograms" to compute feature value frequencies. The values should be booleans indicating the setting. To fully turn off statistics computation pass statistics_config=False. Defaults to None and will compute only descriptive statistics.

Returns

OnDemandFeatureGroup. The on-demand feature group metadata object.

[source]

create_training_dataset#

FeatureStore.create_training_dataset(
    name,
    version=None,
    description="",
    data_format="tfrecords",
    storage_connector=None,
    splits={},
    location="",
    seed=None,
    statistics_config=None,
    label=[],
)

Create a training dataset metadata object.

Lazy

This method is lazy and does not persist any metadata or feature data in the feature store on its own. To materialize the training dataset and save feature data along the metadata in the feature store, call the save() method with a DataFrame or Query.

Data Formats

The feature store currently supports the following data formats for training datasets:

  1. tfrecord
  2. csv
  3. tsv
  4. parquet
  5. avro
  6. orc

Currently not supported petastorm, hdf5 and npy file formats.

Arguments

  • name str: Name of the training dataset to create.
  • version Optional[int]: Version of the training dataset to retrieve, defaults to None and will create the training dataset with incremented version from the last version in the feature store.
  • description Optional[str]: A string describing the contents of the training dataset to improve discoverability for Data Scientists, defaults to empty string "".
  • data_format Optional[str]: The data format used to save the training dataset, defaults to "tfrecords"-format.
  • storage_connector Optional[hsfs.StorageConnector]: Storage connector defining the sink location for the training dataset, defaults to None, and materializes training dataset on HopsFS.
  • splits Optional[Dict[str, float]]: A dictionary defining training dataset splits to be created. Keys in the dictionary define the name of the split as str, values represent percentage of samples in the split as float. Currently, only random splits are supported. Defaults to empty dict{}, creating only a single training dataset without splits.
  • location Optional[str]: Path to complement the sink storage connector with, e.g if the storage connector points to an S3 bucket, this path can be used to define a sub-directory inside the bucket to place the training dataset. Defaults to "", saving the training dataset at the root defined by the storage connector.
  • seed Optional[int]: Optionally, define a seed to create the random splits with, in order to guarantee reproducability, defaults to None.
  • statistics_config Optional[Union[hsfs.StatisticsConfig, bool, dict]]: A configuration object, or a dictionary with keys "enabled" to generally enable descriptive statistics computation for this feature group, "correlations" to turn on feature correlation computation and "histograms" to compute feature value frequencies. The values should be booleans indicating the setting. To fully turn off statistics computation pass statistics_config=False. Defaults to None and will compute only descriptive statistics.
  • label Optional[List[str]]: A list of feature names constituting the prediction label/feature of the training dataset. When replaying a Query during model inference, the label features can be omitted from the feature vector retrieval. Defaults to [], no label.

Returns:

TrainingDataset: The training dataset metadata object.

[source]

get_feature_group#

FeatureStore.get_feature_group(name, version=None)

Get a feature group entity from the feature store.

Getting a feature group from the Feature Store means getting its metadata handle so you can subsequently read the data into a Spark or Pandas DataFrame or use the Query-API to perform joins between feature groups.

Arguments

  • name str: Name of the feature group to get.
  • version Optional[int]: Version of the feature group to retrieve, defaults to None and will return the version=1.

Returns

FeatureGroup: The feature group metadata object.

Raises

  • RestAPIError: If unable to retrieve feature group from the feature store.

[source]

get_on_demand_feature_group#

FeatureStore.get_on_demand_feature_group(name, version=None)

Get a on-demand feature group entity from the feature store.

Getting a on-demand feature group from the Feature Store means getting its metadata handle so you can subsequently read the data into a Spark or Pandas DataFrame or use the Query-API to perform joins between feature groups.

Arguments

  • name str: Name of the on-demand feature group to get.
  • version Optional[int]: Version of the on-demand feature group to retrieve, defaults to None and will return the version=1.

Returns

OnDemandFeatureGroup: The on-demand feature group metadata object.

Raises

  • RestAPIError: If unable to retrieve feature group from the feature store.

[source]

get_online_storage_connector#

FeatureStore.get_online_storage_connector()

Get the storage connector for the Online Feature Store of the respective project's feature store.

The returned storage connector depends on the project that you are connected to.

Returns

StorageConnector. JDBC storage connector to the Online Feature Store.

[source]

get_storage_connector#

FeatureStore.get_storage_connector(name)

Get a previously created storage connector from the feature store.

Storage connectors encapsulate all information needed for the execution engine to read and write to specific storage. This storage can be S3, a JDBC compliant database or the distributed filesystem HOPSFS.

If you want to connect to the online feature store, see the get_online_storage_connector method to get the JDBC connector for the Online Feature Store.

Getting a Storage Connector

sc = fs.get_storage_connector("demo_fs_meb10000_Training_Datasets")

td = fs.create_training_dataset(..., storage_connector=sc, ...)

Arguments

  • name str: Name of the storage connector to retrieve.

Returns

StorageConnector. Storage connector object.

[source]

get_training_dataset#

FeatureStore.get_training_dataset(name, version=None)

Get a training dataset entity from the feature store.

Getting a training dataset from the Feature Store means getting its metadata handle so you can subsequently read the data into a Spark or Pandas DataFrame.

Arguments

  • name str: Name of the training dataset to get.
  • version Optional[int]: Version of the training dataset to retrieve, defaults to None and will return the version=1.

Returns

TrainingDataset: The training dataset metadata object.

Raises

  • RestAPIError: If unable to retrieve feature group from the feature store.

[source]

sql#

FeatureStore.sql(query, dataframe_type="default", online=False)