How to Predict Attendance of Theatre Performances: Discussion at GAMES Seminar

Will the theatre auditorium be full of spectators or will the seats remain empty? Which factors does it depend on? For any theatre the prediction of attendance for specific plays is an important vital task while for academics it is an interesting research work.

Evgeniy Ozhegov and Alina Ozhegova try to solve the problem of predicting the demand for Perm opera and ballet theatre. At the GAMES seminar which took place jointly with research and training group “Empirical demand estimation” the researchers presented their attempt of solving this task. Methodology was the central focus of the seminar.

Within the framework of the research the authors have built a quantile regression model. Using of such complicated econometric framework was justified by heterogeneity of supply. The theatre auditorium is divided into zones which differ, first of all, by the convenience of watching the performance and, secondly, by pricing policy.

Accuracy is the key indicator for making a forecast. In this regard the model is corrected with due consideration of its mistakes – such approach is called “gradient boosting”. At the final stage all plays are divided into groups in accordance with their characteristics, and the model is tested on each group.

The seminar participants discussed basic disadvantages of the model and the ways of splitting original sample into reference and test group, because the model forecast is very sensitive to it.

We would like to note that the first study by Alina Ozhegova (GAMES research assistant and junior participant of the research and training group “Empirical demand estimation”) based on the data of the Opera and ballet theatre was awarded as the best bachelor’s work at the School of economic analysis in 2016.