San Antonio—In this post-Moneyball world of data analytics, it’s an easy stretch to imagine that IBM might use data analytics to help the Orlando Magic win two consecutive games of an NBA playoff series. But what if I tell you that it happened in 1997?
Indeed, the Magic used data mining software created by IBM Research, which matched statistical analysis of the teams against video evidence of their games, according to Inderpal Bhandari, IBM’s global chief data scientist. The software, which Bhandari created in his first job with IBM in the 1990s, told the team one thing: Play two of its backup players more. When the Magic did so, the team won—most of the time.
The point of the story, Bhandari told an audience at the Texas FreshAir Big Data & Data Analytics conference at The University of Texas at San Antonio, is that you not only need insight—the data—but also something that helps you take action on that insight. In this case, that was the video evidence, which validated the success the backup players had against the team’s opponent, the Miami Heat.
“Just imagine if I didn’t have the video to back me up,” Bhandari says. “The video was the next step in the progression to make that decision.”
Data is playing an ever-growing role in the business world, with machine learning software and artificial intelligence systems like IBM’s Watson leading the way. IBM’s Bhandari says that IBM is focused on training more data scientists and engineers in analytics systems, such as open-source computing framework Apache Spark, because it believes the amount of human talent has not kept pace with advances in analytics.
Simply put, there are not enough people who can take action on insight that is provided by today’s data-analytics software, like the video footage of the Magic-Heat games did for the in-game stats, he says.
IBM’s corporate strategy in the current market is on cognitive systems, such as Watson, Bhandari says. The company hopes to use those systems to help businesses in any variety of industries better understand “unstructured data,” such as data that may be based on video, text, or voice information, he says.
“Eighty percent of data is probably unstructured,” Bhandari says. “It’s largely unused in an analysis, decision-making sense.”
To do so, Bhandari says training more data scientists is necessary. IBM has promised to train 1 million data scientists and engineers at its Spark Technology Center in San Francisco, in part working with coding programs such as Galvanize and DataCamp.
After all, human talent is what matters the most, Bhandari says. The Magic’s opponent in the 1997 playoff series, the Miami Heat, was a more talent team and subsequently won the series three games to two, despite IBM’s help.
“Overall, it’s about talent. If you’ve got the talent, this program is going to be helpful,” Bhandari says. “The one team that never used our program was the Chicago Bulls. They never needed our program.”
