An Iterative Model for Predicting Film Attendance
As an important index during film distribution, film attendance is frequently taken into consideration by distribution companies and theater lines when making decisions about budget allocation. Lacking automatic solutions, film attendance is usually estimated by human expertise, which costs many efforts but still cannot achieve satisfactory accuracy. Therefore, it is important to predict film attendance automatically and accurately during film distribution. In this paper, we propose an approach to predicting film attendance of incoming days with film metadata, audience want data, and attendance pattern. An Attendance Iterative Model (AIM) is constructed by iteratively combining random forest based Base Model and SVM based Auxiliary model. The approach has been evaluated with all films released in China in 2015–2016. The result indicates that our model performs well for various films at most times, which MAE maintains within 2–8. Additionally, our iterative model outperforms multi-model with reasonable accuracy and satisfied flexibility of prediction time range.
KeywordsFilm attendance Machine learning Iterative model
- 4.Panaligan, R., Chen, A.: Quantifying movie magic with Google search. Google Whitepaper Ind. Perspect. + User Insights. 1–11 (2013)Google Scholar
- 7.Box, G.E.P., Jenkins, G.M., Reinsel, G.C.: Time Series Analysis - Forecasting and Control, pp. 837–900. Prentice Hall, Englewood Cliff (1994). SFB 373Google Scholar
- 11.Zheng, Y., Yi, X., Li, M., Li, R., Shan, Z., Chang, E., Li, T.: Forecasting fine-grained air quality based on big data. In: Proceedings of the 21th ACM SIGKDD International Conference on Knowledge Discovery and Data Mining (KDD 2015), pp. 2267–2276 (2015)Google Scholar
- 12.Douban Movie. http://movie.douban.com. Accessed 12 Jan 2016