Abstract:
All kinds of factors are taken into account in this paper. Then the least angle regression (LARS) algorithm is applied to implement influence factors selection and the support vector regression (SVR) is employed to fit the total box office returns with the selected main factors. The empirical research shows that the volume, the film format, the releasing time and the genre of a film are the main influence factors of the total box office returns. And LARS-SVR not only avoids overfitting phenomenon of SVR by variable selection, but also retains similar fitting accuracy with LARS and stepwise regression and has far less forecasting error than the latter two models. Finally, on the basis of empirical results, some recommendations are proposed as the reference for the producer, the marketing company and the cinema operator to make decision.
