[snidane]

I'm struggling to understand what the feature store is.

Is it another name for an OLAP or BI cube? Ie. a huge precomputed group by query with rollups.

The only new thing I see is that it combines both historical and recent data. Kinda like an olap cube with lambda architecture.

[willempienaar]

It may be worth reading this article to get an idea of what feature stores are and how they are used https://feast.dev/blog/what-is-a-feature-store/

[Jensson]

A feature store is just a set of features stored somewhere or with the instructions how to compute them. The benefit is that you can easily add new computed features to the store by telling it how it is computed, and then that new feature just works like any other feature. This lets you create a lot of reusable features and signals just by adding the code without touching the data. So you can do code review, add documentation to them and spread the knowledge hose these signals works throughout the organization.

I'm not sure how this relates to OLAP cubes since I am not aware of that term.

[civilized]

SQL also enables stored computation of columns through views. How does a feature store differ from a view? Or is it just a different name for basically the same thing?

[yiksanchan]

I suggest checking out the link from @willempienaar. He is indeed the domain expert since he is building the open source feature store https://feast.dev/ :)

[giovannibonetti]

If I understand it correctly, a feature store is an API that serves data obtained through some sort of (materialized) view. The feature logic can be partially implemented outside the database (which is either stored in memory or on disk).