A Novel Model to Predict a Book's Success in the New York Times Best Sellers List

The popularity of media, such as books and music, has historically been considered difficult to forecast. This popularity is important in determining the success that can be achieved once the media is published. Therefore, we aim to evaluate the extent to which this fact holds true, as we propose that these public opinion trends are quite deterministic in nature. We investigated the important non-textual attributes determining a book’s popularity, including but not limited to a book’s previous ranking in the New York Times Best Sellers list and its popularity in searches. We then constructed two models: a generalized classifier of a successful Best Seller and a predictor of a book’s weekly rank on the New York Times Best Sellers list. The reasonable accuracy of our classification and regression models suggest that book popularity is indeed deterministic. These findings point towards definitive characteristics that can help creators produce successful works.