AI and Analytics in Production

Chapter 3. Artificial Intelligence and Machine Learning in Production

Machine learning and artificial intelligence (AI) have become mainstream processes for many businesses because of their considerable potential to unlock value from big data. In light of this fairly new and increasing interest, we focus this chapter on AI and machine learning as a special example of big data in production. That is not to say that these processes stand apart entirely from what matters for all data-intensive applications in production. The same concerns, goals, and habits of best practice that we’ve already presented apply here, as well, but in several ways AI and machine learning shine a stronger light on some aspects of production. And in some cases, there are special requirements imposed by the special needs of AI and machine learning applications. Our goal is to give you practical pointers that can help make these systems a success in production applications.

AI and machine learning are favorites of ours. We’ve cowritten four books on aspects of these topics (three of them for O’Reilly), and one of us (Ted) has built a number of machine learning systems at several different companies that ranged from recommenders for music, video, or retail sales to detection and prevention of identity theft. We’ve also seen how many MapR customers are using machine learning and AI to extract value from their data in production deployments. From all of this, we’ve figured out some of the things that make a ...