Fall of 2007
Welcome to the Management Science and Operations Management (MSOM) Seminar at the School of Business Administration at the University of Miami.
Here you will find details about the talks given during the Fall of 2007.
Past Talks
- December 14th
Speaker: Ling Wang, Management Department, University of Miami
Title: Joint Price and Due-Date Quotation: Monopolistic and Competitive Cases
Abstract: We consider a joint due-date and price quotation problem in the make-to-order setting under both monopolistic and competitive environments. In the monopolistic model a single firm quotes price and due-date to incoming customers. Customers either accept or reject the joint quote according to a probabilistic acceptance function. We show the firm's optimal policy is to promise the earliest available due-date and offer a price uniquely determined by its current backlog size. In the competitive model two firms compete to win customers by offering a joint quote better than the rival. We provide conditions for the existence of Nash equilibria and characterize due-dates and prices at equilibrium. We identify and explain a seemingly counter-intuitive behavior - "lose-for-now" - whereby the firm with smaller backlog attempts to lose the current customer to the rival. We conduct numerical studies to examine the impact of system parameters on prices and expected profits at equilibrium. We identify two factors that dampen the competition and lead to high prices: backlog imbalance between firms and high utilization.
- December 4th
Speaker: Hernan Awad, Management Science Department, University of Miami
Title: Importance Sampling for Markov Process Expectations
Abstract: We study the use of importance sampling (IS) to estimate a class of expectations of functionals of Markov processes. These expectations are more general than rare event probabilities, and include the expected cumulative discounted reward until hitting a set as a particular case. Experience in the rare-event simulation context shows that characterizing the form of the zero-variance importance distribution can often help in designing an efficient importance sampling estimator. For more general expectations in the Markov setting, however, the zero-variance importance distribution often destroys the Markovian nature of the underlying process, making it difficult to simulate. For the class of expectations we consider, we show that a zero-variance estimator can be constructed by using an importance distribution that preserves the Markovian nature of the underlying process. The importance distribution depends on the solution of a linear system which characterizes the desired expectation. This suggests that, if an approximate solution to the linear system is available, it can be used to construct a good practical importance sampling distribution that preserves the Markovian nature of the underlying process.The first part of the talk will consist of a brief primer on importance sampling techniques.
- November 14th
Speaker: John Liechty, Smeal College of Business, Penn State University
Title: Attribute Level Heterogeneity: Improving Model Based Clustering by Aggregating Information Along Each Dimension
Abstract: In the world of marketing, clustering consumer preferences (commonly referred to as consumer heterogeneity) has played an important role in understanding market structure and allowing practitioners to engage in group and individual level marketing actions. In practice, this clustering has been done using finite mixture models which offers a tool for gaining stylized insights into market segments. Finite mixture models typically cluster 'partworth' utility vectors in P dimensional space - assuming that all of the elements of the partworth vector parameters are the same for each mixture component, a vector finite mixture model. We propose and explore an alternate finite mixture specification, where each element of the partworth vector has its own finite mixture, an attribute finite mixture model. We address a number of statistical issues, such as contrasting the alternate specification with the traditional model and the computational challenges associated with model choice for the attribute mixture model. In addition, we demonstrate that by aggregating information along each of the dimensions of the partworth vector, the attribute model is able to identify a complex structures with less data, when compared to the vector model; leading to a robust approach to segmentation which results in complex market structure that is readily accessible to managers.
- October 26th
Speaker: Amitabh Sinha, Ross School of Business, University of Michigan
Title: Integrated Optimization of Procurement, Processing and Trade of Commodities
Abstract: We consider an integrated optimization problem for a firm involved in procurement, processing and trading of commodities. We first derive optimal policies for a risk-neutral firm, when the processed commodity(ies) are sold using futures instruments. We find that the optimal procurement quantity is governed by a threshold policy, where the threshold is independent of the starting inventory level, and it is optimal to postpone all processing till the last possible period. We extend the model to include risk-averse firms, using a Value-at-Risk constraint on the total expected profits. We show that the optimal procurement quantity for a risk-averse firm is never greater than that for a risk-neutral firm and a risk-averse firm may find it optimal to process and sell the output commodity in earlier periods. We conduct numerical studies to quantify the benefit from integrated decision making and the impact of risk-aversion on expected profits. (For a copy of the paper contact Tallys Yunes.)
About the Speaker: Amitabh Sinha's research interests include logistic network design; social networks and their impact on business innovation; decision making under uncertainty; supply chains; location theory; and product placement. Among other projects, he is currently working on developing models of interactions among employees in corporations, and exploring how the inter-employee social network impacts information flow and innovation in an organization. He is also exploring why, in a competitive market, the first entrant may not always succeed: Along with a colleague, he has found that contrary to popular belief, the market structure could be such that the second entrant always captures greater market-share than the first entrant.
- September 28th
Speaker: Nan Kong, Weldon School of Biomedical Engineering, Purdue University
Title: Issues and Challenges in Optimal Design for Cadaveric Liver Sharing
Abstract: Cadaveric liver transplantation is the only viable therapeutic option for end-stage liver disease patients who have no living donors. However, this type of transplantation is hindered in the United States by donor scarcity and organ viability decay. Given these difficulties, the current U.S. liver transplantation and allocation policy attempts to balance allocation likelihood and geographic proximity by allocating cadaveric livers hierarchically. In this talk, we mainly consider the problem of maximizing the intra-regional transplant efficiency through the design of organ harvesting regions. We formulate the problem as a partitioning problem that clusters organ procurement organizations (OPOs) into regions. We refine the estimate of viability-adjusted transplant quantity to capture the tradeoff between large and small regions. Our partitioning formulation includes too many potential regions to handle explicitly, which leads us to a branch-and-price approach. The pricing problem is an NP-hard nonlinear 0-1 program for which we provide a linear reformulation and design a decomposition heuristic to generate promising columns quickly. Finally we present computational studies that show the benefit of region design and the efficacy of our branch-and-price approach. Our test instances are generated based on recent clinical data. Almost all instances can be solved within a reasonable amount of time and the resulting optimal designs indicate an average viability-adjusted increase of nearly 9% over the current regional configuration. In the remainder of the talk, we present issues and challenges in network configuration and policy design for the U.S. organ transplantation and allocation system.
About the Speaker: Nan Kong is an Assistant Professor in the Weldon School of Biomedical Engineering at Purdue University. He joined Purdue BME in August 2007. Prior to the appointment at Purdue, he was an Assistant Professor in the Department of Industrial and Management Systems Engineering at the University of South Florida, Tampa, FL. He received his Ph.D. in 2006 from the Department of Industrial Engineering at the University of Pittsburgh. His research interests include health systems engineering, data-driven modeling in medical decision making, stochastic optimization, and computational discrete optimization. He received an Honorable Mention for the Dantzig Doctoral Dissertation Award given by the Institute for Operations Research and the Management Sciences (INFORMS) and the 1st Place for the Pritzker Doctoral Dissertation Award given by the Institute of Industrial Engineers (IIE).