Splice Machine Feature Store Debuts

January 20, 2021

Splice Machine has launched the Splice Machine Feature Store. The solution will help more companies operationalize machine learning by reducing the complexity of feature engineering and allow data scientists to make the right decisions based on real-time data.

Despite the hype around AI and machine learning, seven out of 10 executives whose companies had made investments in artificial intelligence reported minimal or no impact from them, according to a 2019 research report from MIT. Those surveyed stated that creating a machine learning model and putting it into operation in an enterprise environment are two very different things.

"The capacity to create, share, explain and reliably reproduce features for a given model is paramount to the success of a data science team," said Monte Zweben, CEO, Splice Machine. "The old way of doing things meant data science operations were simply not scalable. The Splice Machine Feature Store enables you to harness complex analytics in real time and transform real-time data into features, so your models are never uninformed. It also stores feature history making training set creation a single click."

Feature engineering is the most time-consuming and expensive task of the data science life cycle. As companies work to operationalize machine learning, current approaches are not scalable because data science productivity is too low to enable widespread adoption. Simplifying the data science workflow by providing necessary architecture and automating feature serving with feature stores are two of the most important ways to make machine learning easy, accurate and fast at scale.

The Splice Machine Feature Store solves some of the biggest pain points of operationalizing machine learning, including:

Reducing the effort of feature engineering

Helping to solve for governance issues, such as bias, drift, or regulatory oversight

Scaling data science operations

Reducing monetary loss from the creation of inaccurate models

This will help data scientists realize numerous benefits, including:

Achieving faster deployments of AI/ML into production by reusing features and avoiding duplicative feature engineering

Spending 80% less time on feature engineering

Developing more informative models via automatic aggregation of raw data

Gaining predictive accuracy of models with near real-time feature updates and consistent training sets

"As our Clinical Advisor application continues to be adopted by more and more clinicians, there will be increasingly more data at our disposal to help hone and create new features that can further improve our application's precision in identifying the course of disease and improving patient outcomes," said IPH (Innovative Precision Health) Chief Scientific Officer, Mark Gudesblatt, MD. "Splice Machine's feature store makes it possible to leverage features developed by data scientists for a particular disease like Multiple Sclerosis to be leveraged by other data scientists for other diseases like Parkinson’s or Alzheimer's."

"We had this sort of a feature store at Airbnb, but it was limited by the fact that we were largely on HDFS," said Robert Yi, CDO at Dataframe and former Airbnb data scientist. "It enabled users to share features, but it didn't solve the online/offline problem. But the solution can obviously be much more elegant if you start with a more amenable database that can function in realtime. Splice Machine seems to be doing exactly that – ML flow integration, database re-injection, Spark lazy loading, easy deployment, and API-less access."

Zweben added, "The Splice Machine Feature Store is the only one in the market that does not have to synchronize two underlying data engines -- an online store and an offline store, simplifying the architecture with one store, lowering costs, and eliminating latency."