Summary

Insights generated from ML models provide a competitive advantage to a business. However, the process is complex and there is a certain level of discipline that needs to be followed to maximize the return on investment. There are certain core components, such as a feature store, a model registry, a code repo, and a catalog, that are necessary to streamline the ML process, as it is very repetitive and it would be a shame to waste the valuable time of data scientists for tasks that are removed from the use case at hand. The model management aspects cannot be ignored either, because once created, an ML asset is a living, breathing entity that needs care and attention to ensure that it is performing as expected.

In this chapter, we looked at Delta through the lens of an ML practitioner and examined how it adds value to their day-to-day operations on several fronts, including feature engineering and reuse, model training with a unified view of the dataset, model reproducibility...