Back to What We Do

Q&A with Dorian Pyle

Dorian Pyle is the data scientist who helped create Numetrics, a McKinsey solution that analyzes R&D effectiveness. He is based in Miami.

What is your background?

I started consulting on the commercial, business, and industrial applications of advanced analytics, machine learning, and artificial intelligence more than 30 years ago. In the late 1990s and early 2000s, I published books and articles on these topics, which are still in print. Apart from serving clients directly, I have spoken at many conferences, been the chief scientist at an analytics start-up, and am the data scientist behind Numetrics’ analytics capabilities.

Why did you join McKinsey?

When McKinsey acquired Numetrics, I was offered the opportunity, as an expert data scientist, to develop and extend the analytics capabilities of Numetrics as a McKinsey Solution and to take part in the analytics work at the firm. 

Why do clients need advanced analytics techniques today?

The powerful tools of machine learning (along with artificial intelligence and data mining) are the only tools available to assist humans in turning massive data into timely, usable insight and actions. However, the tools are just tools and require skilled human users—just like any other tools we have developed.  Today’s human-in-the-loop machine learning drives timely, effective decision making that was unavailable before now. Our clients must master these skills and technologies to remain competitive.

What is machine learning?

Machine learning is the part of artificial intelligence (itself a part of computer science) dealing with automated pattern recognition, usually in large and noisy data sets. The automated algorithms change themselves (learn) to reflect valid relationships and patterns, and some can recognize novel and changing patterns and relationships—usually far faster than humans. Let’s look at a few examples of machine learning: speech recognition, as in Siri or Cortana, is machine learning. Finding flights is machine learning. So is predicting weather and cancer. This truly is big, exciting stuff, not just boring pattern recognition!

How is it helping clients?

We can see the impact of machine learning in almost all areas of a company’s operations. The most common users are in customer-facing areas such as sales and customer-relationship management. But machine learning also plays a large role in areas you might not expect, like supply chain management, human resource planning, environmental impact mitigation, and so on. As an example, one of my current clients is using machine learning to manage asset risk by predicting maintenance needs on expensive machines.

Where is machine learning headed?

As a professional in a field that partly deals with making predictions, I know how ineffective that endeavor is! But machine learning is a relatively new technology that is maturing very fast.  There will certainly be new algorithms and capabilities developed—but the immediate future is more about the development of the people (the humans who will be in the loop) who can leverage these techniques and make better, faster, smarter decisions.

Published work

Data Preparation for Data Mining, (Morgan Kaufmann, 1999)

Business Modeling and Data Mining, (Morgan Kaufmann, 2003)