Amazon SageMaker Best Practices

Chapter 5: Centralized Feature Repository with Amazon SageMaker Feature Store

Let's begin with the basic questions – what is a feature store and why is it necessary? A feature store is a repository that persists engineered features. A lot of time goes into feature engineering, sometimes involving multi-step data processing pipelines executed over hours of compute time. ML models depend on these engineered features that often come from a variety of data sources. A feature store accelerates this process by reducing repetitive data processing that is required to convert raw data into features. A feature store not only allows you to share engineered features during model-building activities, but also allows consistency in using engineered features ...