According to Gartner's Nick Heudecker, 85% of data science projects fail. This is a staggering ratio of failure that deserves consideration from your organization before undertaking your next data science initiative. In order to keep your team from falling into the chasm of failed data science three key ingredients are required for your next project: events, engineering, and teamwork. This talk will frame those ingredients as the “three pillars of reactive machine learning” and define what each pillar entails.
The first pillar of a successful machine learning project is also one of the foundations of reactive systems: events. Events form a common bridge between the worlds of business, applications and the data analytics that drive machine learning systems, and events are where reactive principles meet machine learning. We’ll cover the most effective technique for modeling the flow of events through your existing systems, and how they will stream into your machine learning pipeline.
The second pillar of reactive machine learning is a well-designed machine learning pipeline based on events. This portion of the talk will outline the most effective techniques for joining events with other sources of complex data, such as a data lake, in order to produce feature vectors for training models and generating actionable insights in near real-time.
Once your machine learning pipeline is in place, we need to achieve project success with the third pillar: teamwork. How does a team of specialists with diverse skill sets align on a common conceptual understanding of the mission? How do we implement the robust processes required in order to take a data science project from inception to production, to business results?
With these three pillars in place on your next data science project, we’re confident that your team can avoid the majority of issues that cause data science initiatives to fail. The Three Pillars of Reactive Machine Learning not only help lead to successful data science outcomes, but they can help redefine your organization as a business based on actionable intelligence. It’s this real-time, actionable intelligence that will give your organization a leg up as the world scrambles to make the promise of data science a reality.
SHARE THIS TALK
Redelastic, Principal Consultant
Kevin Webber has over 18 years of Java development and architecture experience. Before starting the boutique consulting firm RedElastic in 2016, he was both a Developer Advocate and an Enterprise Architect at Lightbend. He was a popular presenter of Lightbend webinars and is a regular speaker on modern enterprise architecture practices.
Redelastic, Chief Scientist
Before joining RedElastic in 2018 Dana Harrington lead teams to successful launches of reactive solutions for projects including a unified e-commerce grocery shopping experience, a third-party marketplace, machine learning based catalog search, and a project leveraging machine learning techniques to help Walmart Canada uncover the key drivers of customer loyalty. Prior to his consulting work, Dana taught post-secondary courses in mathematics, statistics, and software development. Dana has been a functional programming enthusiast for over twenty years.