Department of Decision Sciences Research

Publications 2003 - 2007 | Faculty

Professor Sergio Koreisha and Associate Professor Yue Fang engage in several projects dealing with economic forecasting issues and problems affecting the construction of regression models. Regression analysis, a widely used statistical modeling tool, is concerned with quantifying relationships among variables such as between sales and various marketing promotions and measures of competition. Understanding these relationships help managers make intelligent business decisions, formulate strategies, and develop forecasts that are critical to future planning and control efforts of an organization. Examples of their current projects include:

• Evaluating the Validity of Inferences When Both Heteroscedasticity and Autocorrelation Are Present.
• Should You Trust Tests of Significance in Econometric Models Based on Asymptotically Consistent Methods?
• Estimation and Forecasting of Regression Models with Mis-specified ARCHGARCH Models.
• Trends and Temporal Aggregation in Macroeconomic Time Series.

To complement this research, Assistant Professor Iain Pardoe is developing novel graphical tools to assess the fit of regression models. Traditional "residual analysis" approaches display graphs that can be difficult to use, and that do not provide much insight into how a deficient model might be improved. In contrast the graphs he has been working on, display relationships in the data together with predicted relationships from the model allowing an easy direct comparison. If there is a match, then the model is probably useful, otherwise, the model should be improved - often the graphs can suggest how to achieve this improvement. His current projects include:

• Average predictive comparisons for models with nonlinearity, interactions, and variance components (with A. Gelman)
• Graphical tools for quadratic discriminant analysis (with X. Yin and R. D. Cook)
• A graphical method for assessing the fit of regression variance functions (with R. D. Cook)
• Tools for understanding multilevel (hierarchical) regressions (with A. Gelman)

One highly interesting and useful application of Professor Pardoe's research involves predicting Academy Award Winners. He has a forthcoming paper, "Applying discrete choice models to predict Academy Award winners" in the Journal of the Royal Statistical Society, Series A.

Associate Professor Nagesh Murthy is currently developing normative and empirical models to address strategic and tactical issues at the interface of operations, marketing, and R&D. For example, Professor Murthy is investigating the impact of simulation training on call center agent performance. These models assist firms to gain a better understanding of methods to improve the efficiency and effectiveness of their supply chain decisions.

Professor Murthy also researches in the area of sustainable supply chains as part of his role in the LCB Center for Sustainable Supply Chain Management. He has a forthcoming publication, "Supply chain implications of recycling" in Business Horizons.

Assistant Professor John Goodale's primary research focus is in the area of workforce scheduling in service operations. A typical setting for this decision-making process is a firm that employs customer service representatives in order to satisfy customer demand when it occurs. Many operations have this environment. For example, large financial, telecommunications, and retail firms have call centers that are subject to uncertain customer demand, and one of the important managerial functions is to schedule customer service employees. Goodale and his co-authors have created a holistic approach that forecasts customer demand based on the firm's staffing levels and consumers' utility for waiting time. His current projects include:

Assistant Professor Michael Pangburn studies the effects of product characteristics and their impact on firm operations. One current project, for example, looks at the impact of high-tech products’ obsolescence on capacity and pricing decisions. His other current projects include:

• Capacity Decisions for High-tech Products with Obsolescence (with S. Sundaresan)
• Product Choice with Recourse: Purchases with Returns (with E. Stavrulaki)
• Product Versioning and Timing for Durable Goods. 
• Inventory Disclosure in Online Retailing (with A. Talalayevsky)

As companies have increasingly outsourced part or all of their operations in order to focus on their core competence, a serious capacity allocation problem for the contractors arises due to the decentralized nature of the current practice and the conflicting interests of the parties involved. Motivated by the widely-implemented online capacity booking systems by leading contract manufacturers, Assistant Professor Tolga Aydinliyim conducts research on coordination and competition issues in production planning and supply chain management, with particular emphasis on outsourcing and subcontracting. As opposed to the common approach in supply chain management research, which focuses on coordination at the aggregate inventory level, his approach puts more emphasis on the timeliness of the production activities and the coordination benefits. In a series of papers currently under review at Manufacturing and Service Operations Management and Management Science, Professor Aydinliyim takes an analytical approach in an attempt to answer the following managerial questions: (i) Can significant benefits be achieved as a result of centralized decision making? In other words, are coordination benefits worth the effort to achieve centralization? (ii) Do all parties involved improve their individual performances under centralized control? If not, how should coordination savings be allocated so that individual agents accept the centralized solution? (iii) Can centralization be achieved without centralized control, i.e. does there exist an instrument, e.g. contract, mechanism, priority rule, etc., which makes strategic decision makers act the way they would under centralized control?