Informs Annual Meeting 2017

SB70

INFORMS Houston – 2017

SB70

6 - Learning the LMP-load Coupling from Data: A Support Vector Machine Based Approach Xinbo Geng, Texas A&M.University, 053E Wisenbaker Building, College Station, TX, 77840, United States, gengxbtamu@email.tamu.edu, Le Xie We investigate the fundamental coupling between loads and locational marginal prices (LMPs) in security-constrained economic dispatch (SCED). Theoretical analysis based on multi-parametric programming theory points out the unique one-to-one mapping between load and LMP vectors. Such mapping is depicted by the concept of system pattern region (SPR) and identifying SPRs is the key to understanding the LMP-load coupling. The SPR identification problem is modeled as a classification problem and a SVM based data-driven approach is proposed. It is shown that even without the knowledge of system topology and parameters, the SPRs can be estimated by learning from historical load and price data. 371F Recent Progress in Global Optimization Sponsored: Optimization, Global Optimization Sponsored Session Chair: Guanglin Xu, University of Iowa, Iowa City, IA, 52242, United States, guanglin-xu@uiowa.edu 1 - A Data-driven Distributionally Robust Bound on the Expected Optimal Value of Uncertain Mixed 0-1 Linear Programming Guanglin Xu, The University of Iowa, 108 Pappajohn Business Building, Iowa City, IA, 52242, United States, guanglin-xu@uiowa.edu We study the expected optimal value of a mixed 0-1 linear program with uncertain objective coefficients following a joint distribution. We assume that the true distribution is not known exactly, but a set of independent samples can be observed. Using the Wasserstein metric, we construct an ambiguity set centered at the empirical distribution from the observed samples. The problem of interest is to investigate the bound on the expected optimal value over the Wasserstein ambiguity set. We compare our approach with a moment-based approach from the literature for two applications. Numerical results illustrate the effectiveness of our approach. 2 - Semidefinite Programming for Nash Equilibria in Bimatrix Games Jeffrey Zhang, Princeton University, 206 Lakeside Road, Princeton, NJ, 08540, United States, jeffz@princeton.edu, Amir Ali Ahmadi We are interested in exploring the power of SDP relaxations for the problem of finding an additive -approximate Nash equilibrium for bimatrix games. We show that for zero-sum games our SDP is guaranteed to return a rank-1 solution. Furthermore, we prove that if a rank-2 to our SDP solution is found, then a 5/11- NE can be recovered for any game, or a 1/3-NE for a symmetric game. We propose algorithms based on iterative linearization procedure of smooth nonconvex objective functions which empirically often recover solutions of rank at most two and close to zero. We then show how our SDP approach can address two (NP-hard) problems of economic interest. Finally, we relate our SDP to the Lasserre hierarchy. 3 - Globally Solving Non-convex Quadratic Programs via Linear Integer Programming Techniques Wei Xia, Lehigh University, Bethlehem, PA, 18015, United States, wex213@lehigh.edu, Juan Vera, Luis F.Zuluaga A quadratic program (QP) is a well-studied fundamental NP-hard optimization problem which optimizes a quadratic objective over a set of linear constraints. We reformulate QP as a mixed-integer linear problem (MILP). This is done via the reformulation of QP as a linear complementary problem, and the use of binary variables together with some fundamental results on the solution of perturbed linear systems, to model the complementary constraints. Reformulating non- convex QPs as MILPs provides an advantageous way to obtain global solutions. To illustrate, we compare the performance of our solution approach with the current benchmark global QP solver quadprogBB on a large variety of QP test instances. 4 - Rank Minimization Problem Xin Shen, Rensselaer Polytechnic Institute, 2000 6th Avenue, The problem of minimizing the rank of a matrix subject to constraints can be formulated equivalently as a semidefinite programwith complementarity constraints (SDCMPCC). We investigate calmness of locally optimal solutions to the SDCMPCC formulation and hence show that any locally optimal solution is a KKT point, under an assumption on the original set of constraints. We develop a penalty formulation of the problem. We present calmness results for locally optimal solutions to the penalty formulation. We also develop a proximal alternating linearized minimization (PALM) scheme for the penalty formulation, and investigate the incorporation of a momentum term into the algorithm. SB71 Apt A801, Troy, NY, 12180, United States, shenxinhust@gmail.com, John E.Mitchell

371E Big Data Contributed Session Chair: Xinbo Geng, Texas A&M University, College Station, TX, United States, gengxbtamu@email.tamu.edu 1 - Internet of Things, Big Data, and New Business Models: A Literature Review Régis Delafenestre, Professor, SKEMA Business School, Av. Willy Brandt, Lille, 59777, France, regis.delafenestre@skema.edu, Xavier Brusset With the advent of Big Data and the Internet of Things, e-commerce supply chains require substantial upheavals to remain competitive. New business models will be invented to take advantage of the enormous flow of consumer and Internet of Things data. These business models will remodel supply chains into demand chains which generate the possibility of inventing new ways of creating value for consumers. As a first step, we present the result of a systematic literature review and qualitative content analysis. We leverage the existing knowledge on big data and business models to identify the dimensions of demand-supply alignment and map the drivers, enablers necessary to invent such business models. 2 - An Empirical Study on Workload and Service Lead Times in the Aviation Maintenance Repair Overhaul Industry Busra Keles, University of Miami, 1251 Memorial Drive, Coral Gables, FL, 33146, United States, bkeles@miami.edu, Murat Erkoc, Nazrul Islam Shaikh Maintenance-Repair-Overhaul (MRO) service industries are constantly subject to internal and external factors that result in unsteady workload such as, unanticipated and/or irregular complex tasks, unexpected customer requirements, supplier related delays, and resource variations. To predict service time of an upcoming order and negotiate due dates, one should consider the current and future workload that capture complex scheduling, planning, queuing, and costing parameters through processing big data. For this purpose, we develop a model to identify the factors that influence workload and capture their mappings to service performance, capacity requirements, and budget. 3 - Understanding Patients’ Purchasing Decisions for Online Healthcare Services an Empirical Study Considering Text Form Interaction Data Tong Liu, Phd Candidate, Shanghai Jiao Tong University, No 1954 , Hua Shan Road, Xu Hui District, Shanghai, 200030, China, liutong3231@163.com Now it is of great value to take a look at how patients make purchase decisions for online healthcare services. Some platforms allow patients to interact with physicians in several free chances before they make an order for a paid remote service, so we want to test whether the interaction information contained in those free trails will influence patients’ purchasing decision. We use Bayesian updating process to describe the formulation of patients’ perceived choice utility when they received the feedbacks from the physicians during interaction, where the prior beliefs come from static information contained on webpages and the updating samples are extracted through text mining methods. 4 - Real-time Segmentation of Consumers for Optimized Targeting Chandra Devarkonda, Executive Consultant, IBM, Potomac, MD, United States, chandra.devarkonda@us.ibm.com This paper addresses the issues of providing insightful guidance to retail customer- facing agents be it machine agents (bots) or human agents using data captured in near real time. It considers inherent human decision biases because of brand perceptions, recent buying behavior and influence of networks in buying decisions. Additionally, the solution demonstrates how to capture, ingest, process consumer information almost as it happens and use the outputs to update the understanding of value the consumer brings to the firm. The resulting output demonstrates classification of customers that incorporate both transactional value as well as social value. 5 - A Hybrid Algorithm for Bayesian Network Structure Learning Based on Wrapper Approach Yi Luo, University of Michigan, 3585 Green Brier Blvd., Apt. 95B, Ann Arbor, MI, 48105, United States, Luo1@email.arizona.edu, Daniel L.McShan, Randall K.Ten Haken, Issam El Naqa Bayesian Network (BN) allows for understanding hierarchical relationship between variables and predicting outcome, and it also achieves competitive performance compared to other statistical learning methods. But learning the structure of BNs without restrictive assumption is NP-hard. We developed a novel hybrid learning algorithm to seek stable BNs structures from high dimensional dataset based on wrapper method, and it was used to explore biophysical relationship behind radiotherapy outcome. In addition to having a high prediction performance based on the cross validation, our method can be easily extended to build multi-objective dynamic BNs for personalized adaptive decision support.

60