Skip to main content

News and Media

Open Main MenuClose Main Menu

OSU software predicts a movie's fate before its release

Thursday, December 15, 2005

Oklahoma State University professors have developed a system for predicting the box office success of movies before they hit the theatres, which could revolutionize the industry.

Research by Drs. Ramesh Sharda and Dursun Delen shows their computer-based system can predict exactly how much revenue a movie will generate 37% of the time and provide a prediction accuracy rate of at least 75% most of the time.

The software offers seven types of perameters that are used to determine the revenue range of a movie before its release. Once the revenue range is determined, the movie is classified in one of nine categories from ‘super flops’ that take in less than $1 million to ‘super blockbusters’ that gross more than $200 million.

“All the variables we use are factors you can usually consider as you are deciding whether to make a movie, so we expect this to be a powerful decision aide for potential investors,” Sharda said.

The Spears School of Business Regents Professor of management science and information systems, and his colleague Delen, picked seven factors to help their so-called “neural network” decide on a revenue range for an upcoming movie. The seven variables include the star value of the cast, the movie’s age rating, the time of release against that of competitive movies, the film’s genre, the degree of special effects used, whether it is a sequel or not, and the number of screens it is expected to appear on at its opening.

“The wonder of our system is that it takes each variable that can be either positive or negative alone and joins them together to build a model with solid information,” Sharda said.

Sharda and Delen have been testing their neural network by using data from actual movies. The pair has input data from 834 movies released between 1998 and 2002 to ensure the system’s reliability.  The OSU researchers have been working on the project for seven years.

“Comparison of our neural network to the models proposed in the recent research literature shows that the neural network does a much better job in this setting,” said Sharda, who adds that future plans include expanding the system for use through a website as well as on DVD.

The system is expected to receive a welcome reception from the movie industry, which is in a slump this year compared to last.

Back To Top
SVG directory not found.