Managing Your NCAA Bracket Decisions
Timothy Powers 270003F3FN email@example.com | | Tags:  decision-making decision-management spss crm cognos business-rules analytics watson ibmsoftware predictive baforum business_analytics predictive-analytics business-analytics
0 Comments | 4,693 Visits
Who is in your final four?
If you're one of the millions filling out a bracket this year (all for fun of course), I'm sure you've been asked or have asked that question.
Yes, it's time when the NCAA men's basketball tournament distracts us from our jobs as we maniacally scan the internet and listen to so-called experts hoping to get that edge and finally master the ancient art of bracketology. Sadly, Paul the Octopus passed away recently so that “secret weapon” is no longer viable.
Sure, accurately predicting which teams are in the Final Four is important, but what separates the masters from the novices is predicting the winners/upsets in the early rounds. You can play it safe and pick the higher seeds to win, but that's a silly strategy. Besides, all four top seeds have only advanced to the Final Four once in 30 years. (Sorry President Obama.)
Rely on the data. On Monday, Nate Silver's FiveThirtyEight ran an article entitled, "How We Made Our NCAA Picks," which took an analytical approach to predicting the winners.
Like IBM, he sees the value in analyzing historical data to make informed – and better – decisions.
And let's be honest, everyone is looking for that competitive edge – whether its bragging rights for the brackets, or outmaneuvering the competition in business. The answers are as simple as mining mountains of data to find Key Performance Predictors (KPPs) – those 15-20 data variables that are the most relevant.
KPPs then help guide any organization to build an amazing level of intimacy and knowledge, allowing them to determine how a specific customer is likely to behave at a precise moment in time.
In the NCAA tournament, Nate analyzed the results for all tournament games since 2003 (a total of 512 games) and evaluated which factors best predicted success. As Nate pointed out, "The goal is to have a system that makes good statistical sense and also makes decent basketball sense, as opposed to identifying a bunch of spurious correlations."
Not all data is created equal. In fact, sometimes the correlations you think exist, turn out to be counter-intuitive. That's where KPPs come into play. And, it's why predictive analytics makes good business sense. For instance, one of our insurance customers learned that clients who remove pets from the house prior to a fire are often convicted of claims fraud. And, phases of the moon are a predictive indicator of when crime is likely to occur.
In the NCAA setting, Nate discovered that teams playing games within 50 miles of their campus have a 24-2 record; and, conversely, teams traveling at least 1,000 miles are 121-174.
Does this change the way you think about your bracket?
That's why IBM is "betting" big on predictive analytics. IBM is hoping businesses will realize that picking "winning" customers based on mascots, team colors or flipping a coin is also a silly strategy.
Today, it's better to rely on the data to be told how to take action than making a haphazard decision that could seemingly be based on unnecessary bias (like picking an alma mater such as Boston University over Kansas). Sorry Terriers!
And, what if you could automate all of these decisions?
What if you could determine when a part might fail in a car? Or the right time and conditions to perform surgery? Or when a crime will occur in a specific part of town?
Or, what if a call center agent at a communications service provider could quickly and easily determine which inbound customer calls are the best candidates for an up-sell, cross-sell or retention offer, and then deploy personalized, real-time recommendations that have the greatest likelihood of acceptance by the customer?
Thousands of these types of daily decisions can now be automated and optimized for significant – and measurable – benefit. No longer are the same bad decisions made over and over again.
Sure, it can dominate Jeopardy!, but can it successfully predict an NCAA bracket? That would be "One Shining Moment."