This issue of Cutter Benchmark Review takes a departure from our typical interests. As you know if you closely follow CBR, we focus on technology issues with either a technical or managerial angle. A recent example of the first approach is our July 2007 issue on agile project development (Vol. 7, No. 7), while an example of the second approach is the February 2008 issue on making and using business cases (Vol. 8, No. 2). Given the nature of our interest in information systems and their management, we often produce an issue that is at the crossroads of the technical and managerial arenas. Our issues on the agile enterprise (April 2008, Vol. 8, No. 4), mashups (March 2008, Vol. 8, No. 3), and open innovation (December 2007, Vol. 7, No. 12) represent recent examples.
Not being the group of people to be boxed into a single mold though, we often like to try something new and test your interest. In the past we have done issues solely devoted to a specific class of application or technology, for example, WiMAX (June 2006, Vol. 6, No. 6) or content management systems (April 2006, Vol. 6, No. 4); with this installment of CBR we once again try something new and focus on a topic that is not strictly speaking associated with information technology and information systems management. Specifically, in this issue we look at forecasting and forecasting techniques in the business and organizational setting.
While we are indeed branching out from what most people would consider the IS realm, a closer look suggests that we are indeed not going that far. First, IS professionals are able to collect, store, and manage unprecedented amounts of data about the organization's operations, employees, and customers. The very first issue of CBR that saw me as the editor (October 2005, Vol. 5, No. 10) and many following were based on this premise. With an increasing amount of precise data, often generated automatically by computing devices, organizations have great opportunities to leverage that data through forecasting. Second, many of the modern forecasting techniques are computationally intensive, thus requiring direct and indirect involvement by the information systems function.
Given my limited knowledge of forecasting and forecasting techniques, it was critical to recruit two experts that could create the survey and knowledgeably, yet intelligibly, comment on the results. Our goal here is both to benchmark current organizational forecasting practices and to stretch your thinking about what your organization could be doing with the substantial amounts of data it undoubtedly collects. I have to say that we assembled an outstanding team.
Offering the academic perspective is Monica Adya, Assistant Professor of Management at Marquette University (USA). Monica, with a PhD in information systems, a membership in the International Institute of Forecasters, and a research interest in the use of knowledge-based systems for business forecasting, has the perfect blend of skills to frame this area for us.
For a person able to produce the practitioner perspective I did not have to look far -- just down the hall. Russell Lloyd, my former colleague at the Cornell University School of Hotel Administration (USA), has a wealth of experience in forecasting and specifically in the use of artificial intelligence techniques applied to forecasting problems. Russ has lent his consulting expertise to a number of organizations in and outside of the hospitality industry, including airlines, real estate, and finance.
Monica begins her contribution by broadly framing the domain: "Forecasting is the process of determining what the future holds in order to support organizational decision-making and planning activities." She then shows how complex forecasting can be in the organizational context and discusses how the forecasting function should be structured within the organization. With this introduction in place, Monica discusses how to practically operationalize forecasts as well as the many nuances and decision points that need to be evaluated to develop good forecasts. She points to valuable resources, but if those are not enough, her contact is available for further explanations. Monica concludes with an evaluation of forecasting software and some very valuable guidelines designed to help you improve your forecasting practice.
Russ takes a slightly different approach than typical CBR contributors do. Leveraging his case experience and his practical knowledge, he takes a narrow focus on forecasting approaches based on artificial intelligence and, more specifically, neural networks. His introduction is quite interesting as well as entertaining. In it, Russ makes a powerful point about the difficulty of forecasting accurately in the organizational context. I am no forecasting expert, but I am willing to venture that it is this difficulty that makes many organizations revert to simplistic methods -- or to no attempt at all. After the introduction, Russ focuses on his main interest and details in broad strokes how a neural network works and how it can be used to create powerful forecasts. His valiant explanatory effort notwithstanding, if you want to try and scope out the potential of neural networks for your business, you may be better off giving Russ a call than trying it yourself. This is particularly true in light of the fact that experience in using neural networks often spells the difference between being able to exploit their full potential and making costly mistakes. In my opinion, the most powerful contribution that Russ makes in this issue is the four short cases he produces. They show both the range and potential power of neural nets, and many of you may find similar forecasting situations being the norm in your organization.
This issue takes a departure from our typical focus, but I think that it will be a welcome departure for many of you. Given the amount of available data modern organizations have at their disposal, creatively and correctly thinking about how to extract value from it in terms of forecasting accuracy can further improve the contribution of the information systems function to the bottom line.
