Prerequisites to Use FeatureStore in teradataml Analytic Functions | Teradata Package for Python - Prerequisites to Using FeatureStore - Teradata Package for Python

System Requirements

• Teradata Vantage: Version 17.20 or later
• teradataml: Version 20.00.00.07 or later
• Teradata connection: Always connect to the database using create_context() before initializing a FeatureStore object.

Feature Requirements

The Teradata Feature Store requires temporal database features to be enabled on your Vantage system. This is critical for feature versioning and time-based queries.

Refer to Temporal Table Support, B035-1182.

Roles and Permissions

Feature Store Administrator

Before ingesting the feature values and metadata, the DBA needs to set up the Feature Store repository and grant access to other users on the Feature Store repository. Verify the DBA who is setting up Feature Store has privileges to create database, tables, and triggers on the database.

Setting up Feature store includes creating the following database objects.

Refer to API FeatureStore.setup() to set up Feature Store.

Refer to FeatureStore.grant and FeatureStore.revoke API to grant and revoke access to a user on Feature Store.

Drop privileges are listed as follows. These privileges allow the DBA to remove a feature store as needed.

Refer to FeatureStore.delete() to delete Feature Store.

Feature Store User

The Feature Store User role is a database user who is responsible for ingesting Feature values and creating datasets on the Feature Store. This role also maintains Feature Store metadata.