|
Management Science -- The Big IdeaManagement science practitioners employ a number of statistical/mathematical techniques to help solve business problems and aid with decision making; made easier by the fact that the amount of available data is exploding, the quality of data is improving, and the computational power required to quickly make use of the data is readily available. Development of innovative statistical/mathematical methods developed within the field of management science or adapted from other disciplines is also enabling businesses to understand and address problems more effectively. For instance, how to predict the demand for new products, whether certain customer segments are more profitable than others, and which sales compensation schemes are effective. What We DoThe Management Science research team works with internal constituents to identify business problems that are recurring and significant. Our principal aim is to apply the most useful statistical technique(s) to draw information from numerical data and thereby aid with decision making. Projects are designed to be low cost endeavors and lead to measurable cost savings, revenue growth, or both. Who We AreThe Management Science team includes our principal investigator, Dr. Phillip Yelland, and Renee Stratulate. We are frequently joined by talented interns who share a similar interest in management science and operations research. A List of Recent ProjectsI. Forecasting Blood Build Conversion "Blood builds" are fully configured servers that are manufactured and sent to inventory or distribution prior to receipt of a firm order from the customer. Businesses often initiate blood builds for products with long leadtimes to minimize the cost incurred from a lost sale. This project seeks to forecast the probability that a blood build will not convert to a firm order to guide possible remedial action. II. In Search of the SPARC Advantage: An Exercise in Hedonic RegressionThis project is centered on an examination of Sun's low-end server sales to estimate the price premium associated with the SPARC architecture. The analysis took the form of a series of so-called hedonic regressions to establish a regression relationship between the price paid for Sun Fire server characteristics. One of these characteristics -- which also include main memory size, disk capacity and the like -- is a server's processor type (SPARC or X86), and so by inspection of the estimated regressions, we isolated the price premium associated with the use of the SPARC architecture. We were particularly interested in delineating the course of the SPARC price premium over time to see if price is resilient in the face of X86 product releases. III. A Bayesian Model for Sales Forecasting at Sun MicrosystemsAn accurate short-term forecast of product sales is vital for the smooth operation of modern supply chains, especially where the manufacture of complex products is outsourced internationally. As a vendor of enterprise computing products whose business model has long emphasized extensive outsourcing, Sun Microsystems has a keen interest in the accuracy of its product sales forecasts. Historically, the company has relied on a judgment-based forecasting process, involving its direct sales force, marketing management, channel partners, and so on. Management recognized, however, the need to address the many heuristic and organizational distortions to which judgment-based forecasting procedures are prey. Simply replacing the judgmental forecasts by statistical methods with no judgmental input was unrealistic; short product life cycles and volatile demand demonstrably confounded purely statistical approaches. This project was initiated to develop a forecasting system (currently deployed by the company) that uses Bayesian methods to combine both judgmental and statistical information. IV. Forecasting for Independent Key ComponentsInventories of optional components in discrete manufacturing are often subject to so-called low count demand patterns. Quantities demanded from such inventories in any given period are sufficiently small that it may be unrealistic to forecast them with conventional models based on the normal distribution, and specialized models may be required. Fortunately, the statistical treatment of low count time series has been the focus of much recent research. This project was initiated to apply some of this research to forecasting demands for optional parts at Sun Microsystems. Specifically, we compared the forecast performance of three simple state-space models using demand data obtained from Sun's inventory management records. The models were estimated using Bayesian methods, prdoucing forecasts in the form of full predictive distributions. The accuracy of these probabilistic forecasts was compared using techniques borrowed from the field of meteorology, allowing us to assess the suitability of the candidate models for this type of application. V. Bayesian Forecasting of Parts DemandAs supply chains for high technology products increase in complexity, and as the performance expected of those supply chains also increases, forecasts of parts demand have become indispensable to effective operations management in these markets. Unfortunately, rapid technological change and an abundance of product configurations mean that demand for parts in high-tech is frequently volatile and hard to forecast. This project was undertaken to develop a Bayesian statistical model to forecast parts demand for Sun Microsystems. The model embodies a parametric description of the part life-cycle, allowing it to anticipate changes in demand over time. Furthermore, using hierarchical priors, the model is able to pool demand patterns for a collection of parts, smoothing out idiosyncratic variation and furnishing calibrated forecasts for new parts with little or no demand history. VI. A Graphical Device for Monitoring the Market Reception of a New ProductThe introduction of a new product is at once one of the most difficult and yet most critical activities undertaken by a business. Demand for a new product has wide-ranging ramifications for supply provisions -- such as manufacturing capacity planning and raw materials procurement -- and for marketing mix considerations, including price adjustments, advertising and promotional spending and even product redesign and reengineering (in the event that additional or revised product features become desirable). At the same time, forecasting new product sales remains notoriously problematic, and for a significant period after introduction raw sales figures provide little information concerning the ultimate uptake (or otherwise) of a new product, allowing little opportunity for timely action, should it be necessary. This project was initiated to develop a graphical device that may be used to explore early sales of a new product, furnishing insight into its reception, facilitating projections for long-term sales, and guiding possible remedial actions. The device rests on the observation that the vast majority of "new" products (particularly those introduced by Sun) are actually replacements for or complements to existing products; marketing managers therefore have strong notions as to the markets addressed by these "new" products, in as far as they are closely related to those addressed by existing ones. The device relies on a technique for multivariate statistical analysis known (amongst other names) as correspondence analysis. A List of Technical Papers
| ||||||||||||||||||||||
