From the Stadium to the Classroom
Evan Boyd will teach his first MLDS class in text analytics after building a career turning data output from sports into useful analysis and predictions.
Evan Boyd is lead data scientist at Stats Perform, and this past college football season, he and his colleagues did something even the national champion Ohio State Buckeyes couldn’t — go undefeated.
Stats Perform uses artificial intelligence (AI) to specialize in sports data collection and analysis. Its algorithms correctly picked the winner in all 11 college football playoff games in the first 12-team tournament, an event that ended with the two-loss Buckeyes winning the national championship.
Now, Boyd is preparing for a different challenge — a classroom full of students in Northwestern Engineering's Master of Science in Machine Learning and Data Science (MLDS) program (formerly the MSiA program).
“The main goal is to make sure that the students fully understand the material and can look back on it and say, ‘Oh yeah, I remember this. I can do this,” Boyd said. “In grad school, you really want to hone your skills and master these things.”
The thing Boyd is talking about is a relatively new technology in which AI turns unstructured text data — like words in books, reports, and research papers — into analyzable information that a machine learning (ML) model can probe for hidden truths.
Text analytics is being applied to a rapidly growing number of fields, like turning physicians’ hand-written notes into risk assessments used to guide the insurance industry or analyzing transcripts of customer service calls to develop consumer sentiment reports.
Boyd is well-versed in the deluge of data that comes with the profession. In leading data science for Stats Perform, he has combined the twin passions of his youth — math and baseball. In fact, his current job fulfills the prophecy he made when asked as a youngster to craft a short autobiography.
“Mine was literally, ‘Evan loves math and baseball. He's going to chart statistics for the Major Leagues in 20 years,’” Boyd said. “The fact that it's all kind of come full circle is pretty crazy.”
Boyd’s work involves the mountains of detailed statistics generated by Stats Perform’s sports data arm, Opta. The data comes from not just a sports season, but from a sports game, and sometimes even a single sports play.
He and his team take that data and analyze it to find new patterns and drive predictive models — such as the one that nailed the college football playoffs.
“Many did not value Ohio State as much as our model and criticized it,” he said. “I am especially proud of our model given the backlash on social media when we first released the predictions on our Opta Analyst. This is why more and more people are realizing it is important to factor in data-driven decisions, like the ones we are making at Stats Perform.”
In his seven years with the company, Boyd has delved into the depths of its Opta data for a wide variety of sports, from Major League Baseball to rugby, and just about every sport in between.
The most rewarding part of the work is when Boyd sees or hears the statistical analysis he and his team did repeated by announcers on national sports broadcasts such as the March Madness NCAA Tournament.
“I'm trying to craft different features using the data so it's more interpretable to use to tell a story in-game or post-game,” he said. “It's cool helping a broadcaster to find what we refer to as 'the magic in the detail of sports' and to get stories out there that make their coverage more entertaining.”
Now, Boyd is preparing to bring his skills from the fields and arenas of athletic competition into what he hopes is a memorable MLDS classroom.
“I’d hate to see this course taught and the students memorize a bunch of stuff and then forget about it in a year or so,” he said. “I'm really hoping that they can take the key subjects and go back and use this because it's a skill that can come up daily, depending on the position. It's a very important topic.”
