Movie forecast Guru: A Web-based DSS for Hollywood managers

Abstract

Herein we describe a Web-based DSS to help Hollywood managers make better decisions on important movie characteristics, such as, genre, super stars, technical effects, release time, etc. These parameters are used to build prediction models to classify a movie in one of nine success categories, from a “flop” to a “blockbuster”. The system employs a number of traditional and non-traditional prediction models as distributed independent experts, implemented as Web services. The paper describes the purpose and the architecture of the system, the development environment, the user assessment results, and the lessons learned as they relate to Web-based DSS development.

Introduction

In the motion picture industry, where the results of managerial decisions are measured in millions of dollars, managers are expected to make the best decisions in the shortest possible time. Success (or mere survivability) largely depends on quickly aligning the organizational resources towards changing market conditions in order to meet (and exceed) the actual (and perceived) needs and wants of the consumers. In order to succeed in such an unforgiving environment, managers (and other decision makers) need all the help they can get. Decision support systems (DSS) can provide this much needed help. DSS (recently, also termed as business intelligence systems) are computer technology solutions that can be used to support complex decision-making and problem solving tasks [46]. A Web-based DSS, specifically, is a computerized system that delivers decision support information and decision support tools to managers (decision makers) via a “thin client” Web browser [38].

Prediction of financial success of a movie is arguably the most important piece of information needed by decision makers in the motion picture industry. Knowledge about the main factors affecting the financial success of a movie (and their level of influence) would be of great use in making investment and production related decisions. However, forecasting financial success (box-office receipts) of a particular motion picture is considered to be a rather difficult and challenging problem. To some “… Hollywood is the land of hunches and the wild guesses” [27] due to the uncertainty associated with predicting the product demand. In support of such observations, Jack Valenti, current president and CEO of the Motion Picture Association of America, once mentioned that “… No one can tell you how a movie is going to do in the marketplace… not until the film opens in darkened theatre and sparks fly up between the screen and the audience” [51]. Trade journals and magazines of the motion picture industry have been full of examples, statements, and experiences that support such a claim.

The difficulty associated with the unpredictable nature of the problem domain has intrigued researchers to develop models for understanding and hopefully forecasting the financial success of motion pictures. Litman and Ahn [27] summarize and compare some of the major studies on predicting financial success of motion pictures. Most analysts have tried to predict the box-office receipts of motion pictures after a movie's initial theatrical release [26], [44]. Because they attempt to determine how a movie is going to do based on the early financial figures, the results are not usable to make investment and have had production related decisions, which are to be made during the planning phase. Some studies have attempted to forecast the performance of a movie before it is released but had only limited success. These previous studies, which are either good for predicting the financial success of a movie after its initial theatrical release or are not accurate enough predictors for decision support, leave us with an unsatisfied need for a forecasting system capable of making a prediction prior to a movie's theatrical release. Our research aims to fill this need by developing and embedding an information fusion-based forecasting engine into a Web-based decision support system that could be used by Hollywood managers.

The paper is organized as follows. Section 1.1 gives an overview of Movie Forecast Guru. Section 2 gives a brief review of the related literature in forecasting financial success of motion pictures and Web-based DSS. Section 3 presents our conceptual architecture along with specifics about its implementation and a brief description of the major modeling components. Section 4 illustrates the use of the system by providing a walk-through using an interaction flow diagram. The method used and the results obtained regarding the user evaluation of MFG are also presented in this section. The paper concludes with Section 5 where a summary of the findings along with the limitations and future directions of the research are identified. Appendix A includes detailed screen shots of interaction with MFG.

Fig. 1 illustrates the conceptual architecture of the Movie Forecast Guru (MFG in short) at a very high level. MFG is a Web-based DSS capable of responding to user requests initiated from within a Web browser. Its engine resides in a Web server and is capable of using data (local and remote), models (local and remote) and a knowledge base to carry out its duties: generating financial success predictions and providing sensitivity analysis on the parameters for a variety of movie scenarios generated by the decision makers (investors, producers, distributors and exhibitors). Each of the prediction models used is implemented as a separate Web service, representing an expert available on demand. The core engine can consult each expert, and can present the results of the individual experts as well as a combined forecast to the user. Compiled data from previous performance can also be fed back to the individual models to improve their forecasting performance. The scenarios evaluated by a user are stored in a database for further analysis and/or reuse.

MFG is implemented as a Web-based DSS as opposed to a “desktop application” for a number of reasons:

• Distributed computing—Web-technology enables us to develop the system in such a way that it has a single point of access/entry (front-end), yet provides the means to access a large number of external links (models and data sources) to construct the content in the back-end. Therefore, the complications of the information creation process is hidden from the end user (decision maker) by encapsulating the details within the web server engine and providing the end user only the information they need to make decisions in an understandable multi-media (graphical and dynamic) format.

• Versioning—With the help of a Web-based infrastructure, the end user is likely to have access to the latest version of the MFG system. In contrast, keeping a client application (a desktop application) up to date with the current version would be quite burdensome, since the models are continuously updated as new data and models become available.

• Platform independence—Web-based DSS can be developed independent of the type and nature of the client's computing environment. This allows the development team to spend more time on advancing the underlying methodology of the system as opposed to translating the client application into different versions so that it can run on a wide variety of possible computing platforms. It is especially true in this application where the diversity of computing platforms is evident. The business side of studios may use Windows-based computers, whereas the artistic community may use Macintosh or other graphics intensive platforms.

• Use of models not owned or developed by the system owner—Sophisticated prediction models might be maintained at distant/proprietary locations. The owner of the system might not own the models but have access privileges to use them via some type of a subscription system. With the advent of the Web and its enabling technologies such as the Web services, this kind of computing infrastructure is becoming more and more popular. These external models can also be thought of as human experts. In fact, in the future we plan to add human expert along with the sophisticated analytical models in our “expert” arsenal so that we can provide the decision makes with the most accurate forecast. This class of distributed, integrated infrastructure utilizing multi-expert prediction system is very hard (if not impossible) to implement using traditional desktop applications.

• Facilitating collaboration among stakeholders—The Web-based DSS approach is also capable of supporting multiple decision makers (i.e., stakeholders, namely, investors, producers, distributors, and exhibitors) allowing them to interact with each other using the MFG forecasting engine from distant locations. Such infrastructure provides a desirable platform for group decision-making in arriving at a consensus where each stakeholder could adjust the parameters of their preferred forecasting model of a potential movie project until a consensus is reached or some remedial action is identified among the stakeholders with respect to planning of the movie.