Demand driven operations management in motion picture exhibition

Katherine Goff Inglis

doi:10.32920/ryerson.14644209

Abstract

New opportunities for operational efficiency in movie exhibition exist as a result of recent developments in the industry, such as the mass-scale conversion to digital cinema, the explosion of customer data sources, and the availability of new channels for watching movies. This dissertation begins with an industry overview and discussion of these trends. A review of existing research on forecasting and scheduling problems in movie exhibition is utilized to identify eight factors for decision support systems for operations management in motion picture exhibition. A prototype decision support system (DSS) is constructed using customer loyalty and point of sale data from Cineplex Entertainment. The DSS that is built considers six out of the eight decision support factors that are identified through four modules. The first module projects audience composition for new release movies using loyalty data and identifies a set of the most important movie attributes for each age group. The second module leverages output from the first module with additional data to forecast ticket sales for individual theatre locations. In constructing the second model three different methods (gradient boosted regression, random forest and traditional multiple regression) are tested and the best performing method is utilized in the DSS. The third module applies the attendance forecasts to make labour recommendations. The fourth module uses output from the second module and extends the micro forecast with a concession food demand forecast which is applied to labour recommendations. The DSS is tested empirically in two different ways; firstly forecasted box office sales are compared to actual box office sales to demonstrate that the forecasts being produced are reasonably accurate. Secondly, the labour recommendations from the DSS are compared to the recommendations from the existing DSS at Cineplex and against the theatre manager’s schedules. The DSS performs better than the current schedule and the theatre manager schedule. The labour recommendations with and without the fourth module are compared to demonstrate the incremental value of the concession food module. The study concludes by highlighting further opportunities to extend the system as well as context for practical applications in the motion picture industry.

Highlights

Theatrical Industry Trends and Research ObjectivesMotion picture exhibition is a popular research topic with a number of different perspectives publicized in mainstream media and academic journals
The first section on Audience Composition begins with a high level comparison of the Adjusted R Square values between the 11 models that make up the module
The second subsection on the Box Office Micro Forecasting model is organized in two parts: ‘New Release Forecast’ and ‘Second-Week & Holdover Forecast’

Summary

Introduction

Theatrical Industry Trends and Research ObjectivesMotion picture exhibition is a popular research topic with a number of different perspectives publicized in mainstream media and academic journals. Demand forecasting for movies, a problem in the marketing science domain, has garnered a significant amount of academic interest largely due to the presence of publicly available box office data. This work utilizes customer behaviour and point of sale data to develop a system for demand forecasting and that can be applied in the areas of cinema labour, marketing and advertising scheduling. This work extends the current body of literature on movie exhibition with a five year theatrical exhibition industry overview, a review of recent trends in the digital era of multiplex management and a new integrated approach to demand driven labour recommendations for large cinema chains. The key difference between the multiplex staffing problem and the retail labor scheduling problem is in the demand forecast

Objectives

Methods

Results

Discussion

Conclusion