2015 Informs Annual Meeting

MD56

INFORMS Philadelphia – 2015

4 - The Effect of Supply Base on Ordering Behavior Haresh Gurnani, Professor, Wake Forest University, School of Business and Center for Retail, Winston Salem, NC, 27106,

3 - Productivity Growth and Environmental Efficiency: A Global Malmquist-luenberger Index Analysis Jayanath Ananda, Dr, Central Queensland University, 120 Spencer Street, Melbourne, 3000, Australia, j.ananda@cqu.edu.au, Benjamin Hampf The paper analyses the productivity of the urban water sector using the global Malmquist-Luenberger index while incorporating an undesirable output – greenhouse gas emissions. Findings indicate that the productivity growth of the sector has declined in cumulative terms. The water source, the level of wastewater treatment and production density showed a statistically significant Herbert Lewis, Associate Professor, Stony Brook University, College of Business, Stony Brook, NY, 11794-3775, United States of America, herbert.lewis@stonybrook.edu, Diana Hagedorn, Thomas Sexton Governor Brown of California has directed the State Water Resources Control Board to implement mandatory urban water reductions of 25%. In this paper, we use DEA in 5 of the 9 water use categories, comprising 95% of the state’s water usage, to identify the reductions possible in each county through the elimination of inefficiency. Where the elimination of inefficiency is insufficient to meet the goal, we use a second linear programming model to allocate additional cuts in an equitable manner. MD56 56-Room 109A, CC Spatial Analysis Sponsor: Location Analysis Sponsored Session Chair: Alan Murray, Professor, Drexel University, 3141 Chestnut Street, Philadelphia, PA, 19104, United States of America, amurray@drexel.edu 1 - Transmax 2: An Expanded Transit Route Covering Model Richard Church, Professor, University of California, Santa Barbara, Santa Barbara, CA, 93106, United States of America, rick.church@ucsb.edu, Timothy Niblett Current et al. (1984, 1985) were the first to suggest a routing problem to minimize distance and maximize demand coverage. This problem characterizes the principal goals of transit route design. Since that time there have been a number of formulations, involving elements like route extension and multiple route design. We discusses the TRANSMax model of Curtin and Biba (2011) and based upon that propose a new, expanded formulation called TRANSMax 2, which allows for greater flexibility. 2 - Max Flow with Buyout: Identifying The Minimum Number of Facilities/personnel Required to Meet Demand Blair Sweigart, Operations Research Analyst, US Coast Guard, 216 Maryland Ave, Norfolk, VA, 23504, United States of America, dbsweigart@email.wm.edu The USCG needs to ID deployability of personnel and mission impact. This reduces to a facility location problem: mission demands are demand nodes, personnel are potential facilities, arc capacities are time requirements. Objectives are: determine min. number of personnel needed to meet demand; ID if a person is critical. If not, ID the tradeoff cost. This is a modification of a max-flow algorithm that pushes excess out of the network intelligently to minimize the required number of personnel. influence on the relative efficiency of urban water utilities. 4 - Saving Water in California: using DEA to Allocate Usage Reductions

United States of America, gurnanih@wfu.edu, Karthik Ramachandran, Saibal Ray, Yusen Xia

Previous experimental research in newsvendor problem has mostly focused on ordering from a single supplier; we investigate how the availability of multiple suppliers would influence the order size and allocation decisions with uncertain demand. We also study how the supply base impacts behavioral insights developed under a single supplier setting.

MD54 54-Room 108A, CC

Computational Optimization and Statistical Methods for Big Data Analytics: Applications in Neuroimaging

Cluster: Tutorials Invited Session

Chair: W. Art Chaovalitwongse, University of Washington - Seattle 3900 Northeast Stevens Way, Mechanical Engineering Building, Room G6, Seattle, WA, United States of America, artchao@uw.edu 1 - Computational Optimization and Statistical Methods for Big Data Analytics: Applications in Neuroimaging W. Art Chaovalitwongse, University of Washington - Seattle, 3900 Northeast Stevens Way, Mechanical Engineering Building, Seattle, WA, United States of America, artchao@uw.edu, Shuai Huang This tutorial describes recent advances in computational optimization and statistical methodologies in the emerging research area of Big Data analytics, with a focus on classification, regression and feature selection. We discuss the mathematical and statistical modeling of these problems and provide an application to brain imaging. Analytics of neuroimaging data can provide a unique and often complementary characterization of the underlying neurophysiological process that may be useful in clinical diagnosis of brain diseases. Environmental Application and Computational Aspects of Efficiency and Productivity Analysis Cluster: Data Envelopment Analysis Invited Session Chair: Herbert Lewis, Associate Professor, Stony Brook University, College of Business, Stony Brook, NY, 11794-3775, United States of America, herbert.lewis@stonybrook.edu 1 - An Algebraic Modeling Language Package for Solving Large-Scale Data Envelopment Analysis Problems Wen-Chih Chen, National Chiao Tung University, 1001 Ta Hsueh Rd., Hsinchu, Taiwan - ROC, wenchih@faculty.nctu.edu.tw, Yueh-shan Chung Algebraic modeling languages for mathematical programming provide a flexible and powerful tool for DEA computation. This talk introduces an algebraic modeling language package for solving large-scale DEA problems. While taking the advantage of flexibility and powerful solvers from a modeling language, this package can solve larger-scale DEA problems without limitation on constraints and variables, and with better computational performance. 2 - Frontier Estimation via Penalized Concave Regression Abolfazl Keshvari, Dr., Aalto University School of Business, Runeberginkatu 22-24, Helsinki, 00100, Finland, abolfazl.keshvari@aalto.fi Concave regression is an important tool in estimating a productive efficiency frontier. However, computing this estimator is a very difficult and time consuming task. The computational burden rises very quickly with increasing numbers of observations. We develop an unconstrained quadratic programming (QP) problem to the (monotonic) concave regression, which outperforms the conventional constrained QP problem. Using our approach, we solve a problem with hundreds of observations in some seconds. MD55 55-Room 108B, CC

249