Predicting Oscar winners at the Academy of Motion Picture Arts and Sciences annual Academy Awards show is a longstanding tradition among movie geeks, entertainment industry press, and people with too much time on their hands. Who else likes guessing the Oscars? Data scientists demonstrating the power of their prediction models, that's who. That's the case with Farsite, an analytics division of IT services provider ICC.
The analytics folks at Farsite have posted their annual Oscar night predictions, as well as a brief overview of how their statistical model works. If last year's predictions are any indication, Farsite's data geeks are either very good at movie-award prognostication -- or really lucky.
"We accurately predicted 5 out of 6 winners in the top categories for the 2013 Oscars: Best Picture, Best Actor, Best Actress, Best Supporting Actor, Best Supporting Actress and Best Director. And we're back here in 2014 to do it all again," the company boasts.
Before taking a closer look at Farsite's statistical model, let's run through its predictions for the 2014 Oscars, which take place on Sunday, March 2. Based on current data, the Best Picture winner will be (drumroll, please) 12 Years a Slave.
[Hiring or looking for a big data job? Read Big Data Hiring: 5 Facts From The Field. ]
Matthew McConaughey (Dallas Buyers Club) will earn the Best Actor nod. Cate Blanchett (Blue Jasmine) will take home the Best Actress prize. Alfonso Cuarón (Gravity) will be named Best Director, Jared Leto (Dallas Buyers Club) will win Best Supporting Actor, and Lupita Nyong'o (12 Years a Slave) will get Best Supporting Actress.
To see the latest odds for each nominee, check out Farsite's Oscar forecast page, which updates its predictions throughout the day.
To forecast probabilities for the winners, Farsite's statistical model incorporates more than 40 years of movie industry and Academy Award-related information, the company says. These film factoids include real-time data and numerous variables, including nominees' previous winning performances and nominations, other guild nominations and wins, and the ever-important -- if intangible -- industry buzz.
"Statistically, we know that tracking the other awards will hone our predictions this week before the Oscars. And that's right now!" Farsite writes.
The prediction model factors in voting results from recent awards-season contests, including the Critic's Choice and Golden Globes events, and data from betting lines and prediction markets. Nominees and winners from Producers Guild, Screen Actors Guild, Writers Guild, and Directors Guild voters are incorporated as well. The model also takes into account each nominee's previous nominations and award histories.
Social media buzz might have less impact on Oscar voting, however. Farsite says that although Leo DiCaprio might be getting a lot of Twitter and Facebook love this week for his Best Actor-nominated performance in The Wolf of Wall Street, the Academy's 6,000 voting members will likely choose McConaughey. That's a prediction based in part on voting results of earlier awards.
"Our predictions are far more than lucky guesses. Most people are surprised to hear that the same sophisticated predictive modeling we use in industries like retail and healthcare can predict Oscar winners quite accurately," said ICC chief science officer Ryan McClarren in a statement.
Will Farsite's predictions ring true this year? Or will its big data statistical model flop like The Lone Ranger did last summer? The envelope, please.
Engage with Oracle president Mark Hurd, NFL CIO Michelle McKenna-Doyle, General Motors CIO Randy Mott, Box founder Aaron Levie, UPMC CIO Dan Drawbaugh, GE Power CIO Jim Fowler, and other leaders of the Digital Business movement at the InformationWeek Conference and Elite 100 Awards Ceremony, to be held in conjunction with Interop in Las Vegas, March 31 to April 1, 2014. See the full agenda here.Jeff Bertolucci is a technology journalist in Los Angeles who writes mostly for Kiplinger's Personal Finance, The Saturday Evening Post, and InformationWeek. View Full Bio