You need data architecture flexible enough to handle random business requests for a logical or even physical data warehouse. A logical data warehouse provides connections to all data sources irrespective of where the data resides. It connects, transforms, and provides a unique access layer for all users, which eliminates miscommunication and over preparation of data because all data is quickly available, connected, prepared, and curated.
It’s important for data analysts and developers to see the data pipeline as a whole because it helps answer questions on where data is coming from, what transformations are happening, what views are being created, and what has been published in a logical or physical data warehouse. Machine learning models should be part of the data flow detailing in and out data streams.