The same technology that is used to combat crime, reduce customer churn, identify part failures in manufacturing and reduce the mortality rates of cardiac victims, is being served up this year at the U.S. Open Tennis Championships.
As John McEnroe would exclaim, rather hysterically, “You cannot be serious!?”
Predictive analytics technology is providing in-depth analysis of all the players in this year’s tournament and will identify the “Keys to the Match” that must be met in order to win each particular match.
IBM PointStream delivers state-of-the-art visualization capabilities that enable users to interact with live match data, providing a more enjoyable viewing experience, and allowing fans to get closer with their favorite players.
Powered by IBM predictive analytics technology (specifically IBM SPSS Modeler data mining & text analytics software), IBM PointStream analyzes more than five years worth of Grand Slam tennis data (39 million data points) to find patterns and styles for players when they win.
Prior to each match, the IBM “Keys to the Match” system will run an analysis of both competitors’ historic head-to-head statistics, as well as stats against comparable play styles, to determine the keys to winning the match. If that player can succeed in accomplishing these three items, statistically speaking, they should be victorious.
In other words, it will find the value of a particular statistic (such as the percentage of successful first serves, better lob shots, or having a stronger backhand) that will increase the player’s chances of winning a set. These are called Key Performance Predictors. By identifying the tendencies of each player, IBM PointStream determines if the strength of one player can be used to exploit an opponent’s weakness.
For instance, used in a commercial setting, an insurance companies use predictive analytics to identify the keys to identify likely cases of fraud. If a customer calls in to report an accident, but there were no other parties involved and the police were not notified, then that scenario has a high probability of being fraudulent.
IBM PointStream, analyzes previous head-to-head matches between two players, but if those players never played each other before or if they only met once or twice, there is hardly enough data for statistical inference.
To enhance the analysis, each player is also classified by the style of tennis that he or she plays, and all players are grouped by their ranking and results within each of the tennis styles. This allows the system to not just consider any direct head-to-head matches, but also to include matches between players, who have similar tennis styles and results.
The individual “Keys to the Match” are automatically generated for each player as soon as the schedule for the next day of the tournament is announced and they are then made available through IBM PointStream once the match begins.
For more than 20 years, IBM has been working with the U.S. Open to bring a better tennis tournament experience to millions of fans worldwide. And for fans following the tournament online, they can now become even more involved.
The power of predictive analytics…
Game. Set. Match.