Forecasting Cinema Attendance at the Movie Show Level: Evidence from Poland

Abstract

Background: Cinema programmes are set in advance (usually with a weekly frequency), which motivates us to investigate the short-term forecasting of attendance. In the literature on the cinema industry, the issue of attendance forecasting has gained less research attention compared to modelling the aggregate performance of movies. Furthermore, unlike most existing studies, we use data on attendance at the individual show level (179,103 shows) rather than aggregate box office sales. Objectives: In the paper, we evaluate short-term forecasting models of cinema attendance. The main purpose of the study is to find the factors that are useful in forecasting cinema attendance at the individual show level (i.e., the number of tickets sold for a particular movie, time and cinema). Methods/Approach: We apply several linear regression models, estimated for each recursive sample, to produce one-week ahead forecasts of the attendance. We then rank the models based on the out-of-sample fit. Results: The results show that the best performing models are those that include cinema- and region-specific variables, in addition to movie parameters (e.g., genre, age classification) or title popularity. Conclusions: Regression models using a wide set of variables (cinema- and region-specific variables, movie features, title popularity) may be successfully applied for predicting individual cinema shows attendance in Poland.

This work is licensed under a Creative Commons Attribution-NonCommercial 4.0 International License.