What is feature storage?

What is feature storage?

Feature storage is an overlay to conventional big data systems that automatically extracts features, not data, from each of the underlying data sources or data lakes and stores them in one centralized feature storage platform. The feature storage platform maintains up-to-the-millisecond data updates with little to no upfront data preparation. This is achieved by reducing the dimensionality of the original data, effectively collapsing conventional data models (such as relational or star schemas) into a highly-optimized format that is natively predisposed for real-time analytics and AI. The feature storage platform then serves feature vectors for training and production purposes and allows for the re-use and sharing of features inside and outside of an organization. Feature storage is typically implemented and managed by data engineers and provides data scientists, ML researchers and application developers a single access point to derive insights, predictions, and real-time decisions from big data. Implementing a feature storage platform allows companies to graduate from reporting and explaining with human-scale data to predicting and prescribing real-time business outcomes on all data.