Prerequisite: None
This session will include a moderated Q&A featuring questions from the live audience.
It is commonly observed that only a small fraction of machine learning (ML) models are deployed. Often the model building is successful—the models meet numerous objective measures of success—yet they are never used. Successful deployment requires ML models to be fully incorporated into the day-to-day practices of the business through automation.
The incorporation of analytics into a business process is a unique kind of business automation. What makes it different from robotic process automation (RPA), for instance, is the presence of a machine learning model. However, AutoML will not help with deployment. Although valuable, AutoML facilitates the building of ML models, not their integration into business processes.
There are three keys to successful ML deployment projects. They need to deliver a clearly defined prediction, help the business make "micro-decisions," and operate in situations where advance knowledge offers a measurable advantage.
In this session, Keith McCormick will define the characteristics of a successful machine learning deployment, discuss the three keys to successful deployment, and show you how to select machine learning projects strategically and steward them to a successful conclusion.