Author: Lulu Liu

Why are enterprise feature stores empty? The notion that “data is the new oil” has existed for a while. The analogy, however, is more appropriate than people often consider. Crude oil serves few concrete purposes in its raw state. It must be refined and processed before gasoline, for example, can power our vehicles. In the same fashion, raw data has little value to an organization. Like crude oil, we must refine, cleanse, transform, and often combine it with other data elements to elicit business insights and value.  In the case of machine learning, the “gasoline” is what is known as “features.” Easy access to diverse and high-performance features is critical for successful machine learning projects. While feature engineering has been around as long as data scientists have built machine learning models, feature stores are a relatively new concept. A Feature store is a machine learning-specific system used to centralize storage,…