Three years ago, we challenged our analytics team to build a model that would predict the Oscar winners for the 87th Academy Awards.  They responded by developing models that correctly predicted the winners of all four major categories, Best Picture, Best Director, Best Actor, and Best Actress.

 

Last year, they went 3 for 4 missing on Best Picture.  This year, we’re at it again with the following predictions for the 89th Academy Awards ceremony this Sunday, February 26, 2017.

 

Best Picture = La La Land

Model probability of winning – 79-90%

Las Vegas probability – 87%

Key Indicators:

Winning best picture at Producer’s Guild Awards, IMDB user rating, Total number of nominations

 

Best Director = Damien Chazelle (La La Land)

Model probability of winning – 75-98%

Las Vegas probability – 93%

Key Indicators:

Winning best director at Director’s Guild Awards, Total Oscar nominations, Winning best director at Golden Globes

 

Best Actor = Casey Affleck (Manchester by the Sea)

Model probability of winning – 63-88%

Las Vegas probability – 62%

Key Indicators:

Winning best Actor (drama) at Golden Globes, Winning best Actor Screen Actors’ Guild, Age of Actor in release year

 

Best Actress = Emma Stone (La La Land)

Model probability of winning – 47-76%

Las Vegas probability – 83%

Key Indicators:

Age of Actress in release year, Winning best Actress at Golden Globes, Total nominations for picture

 

Yes, our computers love La La Land too, and they haven’t seen anything but the data.  There has been some backlash since its domination of the Golden Globes, and with Hollywood taken to task last year over a lack of diversity, don’t count out Moonlight, Fences, or Hidden Figures in the best picture category.  We also wouldn’t be surprised to see Denzel Washington take the best actor award.

 

The Science

To generate our predictions, we used three machine-learning models commonly used to classify unknown quantities based on known data.  In this case, we are classifying the Oscar nominees as winners or losers based on what we know about the winners from previous years.  We encounter these types of models all the time.  They help us find movies we’d like to watch on Netflix, or books we’d like to read on Amazon.  They also play a role in medicine.  A similar classification model developed in part at the Children’s Hospital of Philadelphia helps to identify babies most likely to develop autism from MRI images of the babies’ brains.  You can find more information on that study here.

Data Sources:

Academy Awards DatabaseInternet Movie Database, Roger Ebert ReviewsScreen Actors GuildProducers Guild, Directors Guild

 

References:

Pardoe, I. and Simonton D. K. (2007) Applying Discrete Choice Models to Predict Academy Award Winners, J. R. Statist. Soc. A(2008), 171, Part 2, pp. 375-394