“Machine Learning” is a trendy term being kicked around (pun intended) by fraud fighters around the world. In fact, Machine Learning is such a popular term that it is becoming a staple in buzzword bingo games.

Here’s a little secret about machine learning… many of the people who talk about it have no real idea what it means!

If you are anything like me, you’ve heard others talk about machine learning. Or, you’ve read an article or blog post about it. Your gut tells you that it’s important, and it might help you do your job better. You’re curious and want to know more. But, you’ve got a day job that consumes well more than 8 hours a day.

Enter the most-watched sporting event on the planet.

What does the global soccer tournament have to do with machine learning? Indulge me for a moment…

Soccer fans are intensely passionate. We follow our teams religiously. We rejoice at wins. We are devastated by losses. And we debate constantly about who will win the tournament.

One intense fan – Andreas Groll – happens to be a math professor. He is such a passionate soccer fan that he developed a mathematical algorithm to predict who will win the big tournament. And yes, you guessed it… it’s a machine learning algorithm.

So who’s gonna win? Dr. Groll’s fancy machine learning algorithm predicts Spain will take the tournament.

As fraud fighters, what can we learn from Dr. Groll’s algorithm? Without having to understand the calculations behind it?!!

1. Machine Learning is challenging the conventional wisdom… and that’s good. Betting odds are a great way to measure public opinion on who will win a sporting event. In Las Vegas, the current favorite to win is Brazil, at 7-2 odds. Remember, though, that the algorithm picked Spain as the winner… not a controversial choice, but one that is counter to the 13-2 odds for Spain assigned by bookmakers. Lesson 1: Machine learning can challenge convention wisdom in a positive way.
2. The ML algorithm tells us WHY it predicts a victory. Dr. Groll analyzed hundreds of potential factors that people think might make a difference – from home field advantage and distance traveled to GDP and population of a country. Machine learning helped to filter out the noisy debate. So what matters? Rankings matter a great deal. So does a country’s GDP. And the number of Champions League players on each side. What doesn’t matter? The nationality of the team manager. Or a country’s population. Lesson 2: Machine learning gives you transparency into why it predicts any given outcome – giving you insight into what truly matters AND what doesn’t matter.
3. The ML algorithm gives us a nuanced analysis that allows us to make a better human judgment. There’s a twist in Dr. Groll’s algorithm. It gives Spain a 73% chance of reaching the quarterfinals. By contrast, it gives Germany only a 53% chance of surviving its group matches. But, if Germany does manage to make it to the quarterfinals, the ML algorithm predicts Germany – not Spain – to be the winner. In short, the ML algorithm thinks Germany has tough matches early in the tournament. If it survives those matches, the chances of a repeat victory for Germany are high. Lesson 3: Machine learning gives you tremendous insight, helping to understand the nuances of a very complex challenge.

To get out in front of fraudsters, you need to get deep insight into complex challenges and understand what matters most and least in your data, and that cannot be achieved with conventional wisdom. You need precisely the things machine learning can deliver.

You still may not fully understand machine learning. And that’s perfectly OK. This algorithm tells you all you need to know -- that ML gives you better and faster ways of understanding your data.

Oh… and one minor detail I’d like to point out… like his ML algorithm, Dr. Groll is from Germany.

Ben Wright, Rodney Carson, and John Maynard contributed to this post.

Tags
Share

Global Leader - Fraud & Integrity

Shaun Barry is a renowned expert in fraud and integrity, with a specific focus on government. Shaun has worked for and with federal, state, and local governments around the world for over twenty (20) years to foster innovative and efficient business processes through technology. He specializes in tax & revenue, healthcare, social benefits, and motor vehicle functions. Shaun holds a Bachelor of Arts degree in American Studies from the University of Notre Dame and a Master of Public Policy degree from Duke University.