2015 Informs Annual Meeting

WC30

INFORMS Philadelphia – 2015

WC29 29-Room 406, Marriott Decision Analytics Applications in the Media Industry Sponsor: Analytics Sponsored Session Chair: J. Antonio Carbajal, Sr Operations Research Analyst, Turner Broadcasting System, Inc., 1050 Techwood Dr., Atlanta, GA, United States of America, Antonio.Carbajal@turner.com 1 - Commercial Time Management Applications in the Media Industry J. Antonio Carbajal, Sr Operations Research Analyst, Turner Broadcasting System, Inc., 1050 Techwood Dr., Atlanta, GA, United States of America, Antonio.Carbajal@turner.com, W. Chaar Advertisement-based networks face the challenge of optimally allocating their commercial time capacity to meet audience levels guaranteed to their advertisers. This presentation introduces two systems that support this objective: 1) A spot scheduling engine that maximizes deal delivery while honoring several placement constraints and 2) A make-goods allocation engine that maximizes the liability inventory value and meets advertiser requirements of selling title mix and weekly unit distribution. 2 - Competitive Audience Estimation W. Chaar, Analytics, Data & Decision Sciences, VP, 106 N Denton Rd, 210, Dallas, TX, 75019, United States of America, wscemail2002@yahoo.com, J. Antonio Carbajal, Peter Williams A consumer choice modeling approach is developed and used to model audience behavior across competing media contents. The methodology demonstrates the capability of the approach to generate accurate predictive results of audiences. 3 - Television Programming Optimization in the 24-hour Cable Network Environment Peter Williams, Sr. Operations Research Analyst, Turner Broadcasting System, Inc., 1050 Techwood Dr., Atlanta, GA, 30308, United States of America, peter.williams@turner.com, W. Chaar, J. Antonio Carbajal A cable network programmer’s main objective is to create program schedules that attract viewers. The mix of available programs and various scheduling constraints introduce complexity to organizing a 24-hour program schedule. With Nielsen data, the authors develop a model to predict ratings using features of a program schedule. Using this model, TV program scheduling is formulated as an optimization problem with the objective of maximizing ratings given program airtime constraints. WC30 30-Room 407, Marriott Information Systems II Contributed Session Chair: Harris Kyriakou, Stevens Institute of Technology, 1 Castle Point on Hudson, School of Business 642, Hoboken, NJ, 07030, United States of America, ckyriako@stevens.edu 1 - Cooperate to Dominate?: Empirical Analysis of Cooperation Decisions on Technological Innovation Ashish Kumar Jha, Doctoral Student, Indian Institute of Management Calcutta, Diamond Harbor Road, Joka, Kolkata, WB, 700104, India, ashishkj11@iimcal.ac.in, Indranil Bose We analyze the importance of cooperation factors and cooperation partners for product and process innovation for Chinese firms. We also test the relationship between product and process innovation which leads to confirmation of our hypothesis that continuous process innovation has a rub off effect on product innovation and product innovation increases but vice versa is not true. We arrive at the conclusions based on analysis of innovation data of over 900 Chinese firms. 2 - Commercial Companies’ Open Source Strategies – When Should Competitive Companies Support the Same Operation Ying Liu, Xi’an Jiaotong University, No.28 Xianning Road, Beilin District, Xi’an, China, liuyingleyna@stu.xjtu.edu.cn Nowadays many firms have been involved with open source software through various ways, such as contributing code and providing support to open source community. They do this for profit, mostly through complementary product or service. How much effort should they put into open source project and does their competitor’s contribution affect their behavior? How would they design additive product? Does free-riding problem really hurt and which kind of license can mitigate it?

4 - P-center and P-dispersion Problems: A Bi-criteria Analysis Golbarg Kazemi Tutunchi, North Carolina State University, 1916 Trexler Ct, Raleigh, NC, 27606, United States of America, gkazemi@ncsu.edu, Yahya Fathi We consider the p-center and the p-dispersion problems in the context of a bi- criteria location analysis. We discuss a mathematical programming approach to obtain a non-dominated point with respect to these two objectives, and an exact method to obtain the corresponding non-dominated frontier. Through a computational experiment we demonstrate the effectiveness of this approach for different instances of the problem. 5 - A Cut and Branch Approach for a Class of Bi-Objective Combinatorial Optimization Problems Sune Lauth Gadegaard, PhD fellow, Department of Economics and Business, Aarhus University, Fuglesangs Allé 4, Bygning In this talk we discuss a cut and branch approach for bi-objective optimization problems with integer outcome vectors. The approach improves the LP-relaxation by adding cuts at each extreme point of the efficient frontier of the LP-relaxation. After improving the LP-relaxation, we propose a branch and bound algorithm which alternately branches in objective space and in decision space. Experimental results for the single source capacitated facility location problem is reported. 2622, lokale 315a, Aarhus V, 8210, Denmark, sgadegaard@econ.au.dk, Matthias Ehrgott Chair: Yasaman Kazemi, North Dakota State Univesrity, NDSU Dept. 2880, P.O. Box 6050, Fargo, ND, 58108, United States of America, yasaman.kazemi@ndsu.edu 1 - Multistage Stochastic Programming under Endogenous and Exogenous Uncertainties Robert M. Apap, Carnegie Mellon University, 5000 Forbes Avenue, Pittsburgh, PA, 15213, United States of America, rapap@andrew.cmu.edu, Ignacio E. Grossmann We address the modeling and solution of mixed-integer linear multistage stochastic programming problems involving endogenous and exogenous uncertain parameters. We propose a composite scenario tree that contains both types of parameters. We then present new theoretical model-reduction properties to drastically reduce the number of non-anticipativity constraints. Lagrangean decomposition and a sequential scenario decomposition heuristic are used to solve large-scale instances. 2 - Stochastic Optimization of Downstream Petroleum Supply Chain under Uncertainties Yasaman Kazemi, North Dakota State Univesrity, NDSU Dept. 2880, P.O. Box 6050, Fargo, ND, 58108, United States of America, yasaman.kazemi@ndsu.edu, Joseph Szmerekovsky An integrated multi-product multi-echelon and multi-mode supply chain design is studied in the downstream petroleum supply chain under the risk of facility disruptions. Two-stage stochastic model is proposed to minimize the expected total costs by optimizing the strategic and tactical decisions and selecting appropriate mitigation strategies. In addition, Geographic Information System (GIS) is used to locate facilities, obtain realistic transportation data, and to visualize the process. 3 - Risk-averse Wind Power Investment through Unified Stochastic and Robust Optimization Kun Zhao, Department of Industrial and Systems Engineering The University of Florida, FL, United States of America, zhaokunzk@ufl.edu, Yongpei Guan A unified stochastic and robust optimization approach is proposed for the wind power investment problems. Two models are developed: a short-term risk-averse wind power investment problem for market participants, and a long-term simultaneous wind power and transmission investment problem for vertically integrated utilities. By reformulation, tractable formulations can be obtained and solved in finite time. Computational results are shown for the effectiveness of our proposed model and solution approaches. WC28 28-Room 405, Marriott Optimization Stochastic V Contributed Session

435