Global Analytics Competition Demonstrates How Worldwide Collaboration Yields Innovative Business Solutions

September 22, 2009

NEW YORK, Sept. 22 /PRNewswire/ — At the photo-finish end of the largest and most prestigious global analytics competition ever, the Netflix Prize, The Ensemble applauds BellKor Pragmatic Chaos (“BellKor”) on its Grand Prize win, and thanks Netflix for its sponsorship of a contest that demonstrates the true power of crowd sourcing and open collaboration in the arenas of machine learning and advanced analytics. The Ensemble and BellKor were the only two teams out of more than 41,000 entrants who qualified for the grand prize by improving the performance of Netflix’s movie recommender program by over 10%.

The Ensemble, a collaboration of three groups – Opera Solutions, Vandelay Industries !, and Grand Prize Team – ended at #1 on the Leaderboard, and was second place finisher in the overall contest.

Ultimately, it was only timing that determined the winner. Each team’s model yielded the exact same 10.6% improvement against the dataset used to determine the final winner. But BellKor submitted its entry 20 minutes before The Ensemble, so under the contest rules, it won.

Although members of The Ensemble made numerous analytic breakthroughs throughout the two-and-a-half-year contest, perhaps the most profound ones lie in understanding the power of group collaboration in solving very difficult analytic problems. “The two qualifiers in the contest are consortia teams,” explains Arnab Gupta, CEO of Opera Solutions. “In fact, if you look at the Top 10, they are merely variations of The Ensemble and BellKor. Individually, we could not crack this problem; only when we joined forces could we get to the top of the Leaderboard. This model of collaboration, I am convinced, is the future of complex problem solving.”

The Ensemble is itself a model of collaboration across borders and disciplines, with team members hailing from more than 10 countries and representing a wide range of academic and professional areas. It was this diversity, many members are convinced, that led to its success. “This competition brought together great talent from around the world,” commented Jacob Spoelstra, Vice President, Analytics, Opera Solutions. “We each thought about the problem differently, and these differences resulted in dramatic improvement when all the teams combined their models during the last month.”

The Ensemble was well-prepared for collaboration. Greg McAlpin, the founder of Vandelay Industries ! whom many team members view as the “glue” for the broader group, recognized early this year that no single competitor could hope to win the contest and began creating the processes and assembling the technology to support rapid model blending. “The Ensemble was able to accomplish an almost impossible feat in an extremely short period of time. This was only possible because of months of work and planning that had gone into creating an environment where a large number of teams would be able to share their work.”

About The Ensemble

The Ensemble is comprised of many disciplines and geographies. Members hail from around the world, including Australia (Mike Linacre), the United States (Aron Miller, Joe Sill, David Lin), Canada (Bo Yang), China (Xiang Liang, Bruce Deng, Peng Zhou), Hungary (Gabor Takacs, Domonkos Tikk, Bottyan Nemeth, Istvan Pilaszy), Poland (Wojtek Kulik), Greece (Nicholas Ampazis, George Tsagas), the Netherlands (Edward de Grijs, Willem Maestrom), India (Priyanka Rastogi, Arvind Gangadhar) and Israel (Dan Nabutovsky). The team includes experts in machine learning (Bill Bame, David Purdy, Lester Mackey); artificial intelligence (Larry Ya Luo, Craig Carmichael); engineering (Ces Bertino, Chris Hefele); biology (Jeffrey Howbert); and computer science (David Weiss).

About Opera Solutions

Opera Solutions (www.operasolutions.com) is a technology-powered consulting firm helping the world’s premier companies quickly achieve sustained profit growth. Founded in 2004, the firm delivers results through its proprietary “Total Data, Total Insights, Rapid Action” approach. Opera’s data management capabilities, advanced analytics, and commitment to end-to-end execution have unleashed billions of dollars in profit for clients in financial services, media, consumer goods, hospitality, and telecom sectors around the world. The company is actively applying the learnings of the Netflix Prize competition in a wide range of areas, including credit granting, fraud and risk mitigation, marketing, pricing, cross-selling, portfolio valuation, and trading.

SOURCE Opera Solutions

Source: newswire

comments powered by Disqus