Audience Ratings Prediction of TV Dramas Based on the Cast and Their Popularity

Television viewership ratings represent the popularity of TV programs and are an important indicator of the revenue generated by broadcasters from advertisements as well as other sources. Although higher ratings for a given program are beneficial for both broadcasters and advertisers, little is known about the factors that make programs more attractive to viewers. In this paper, we propose a method to predict the ratings of TV dramas before they are broadcasted by considering the cast and the staff involved with them. In order to consider the popularity of actors, we consider the number of hits received by their Wikipedia pages, and tweets related to them on Twitter. We tested our proposed method using a collection of 678 TV dramas. The experimental results showed that audience ratings can be predicted with a correlation coefficient of r = 0.845. The effect of each factor was also investigated to reveal the components significant to audience ratings.