Sundance successes predicted by Bayesian filters

The LA Times is reporting on an anti-spam company that has used Bayesian filtering to predict, with 81% accuracy, which films at the Sundance Film Festival will financial successes. The company's enginneers realized that "determining whether a movie is good or bad could be similar to determining whether e-mail is spam or not" and used the last 10 years of the festival to determine the predictors of success or failure. Among the predictors are the number of producers (more is better) and the words in the blurbs (good: romance, death, human; bad: Africa, riveting, vision). The project, called Deconstructing Sundance, has its own web site where you can check out their predictions for this year's films.

Tags: bayesian, deconstructing sundance, film, movies, news, spam, sundance, unspam