Informs Annual Meeting Phoenix 2018

INFORMS Phoenix – 2018

TD71

n TD70 West Bldg 106B

n TD71 West Bldg 106C Joint Session ICS/Practice Curated: Networks and Their Applications Sponsored: Computing Sponsored Session Chair: Hamidreza Validi, Oklahoma State University, Stillwater, OK, 74075, United States 1 - The Most Betweenness Central Clique Problem Foad Mahdavi Pajouh, University of Massachusetts Boston, Boston, MA, 02125, United States, Maciej Rysz, Eduardo Pasiliao This talk addresses the most betweenness-central clique problem, which is to find a clique of maximum betweenness centrality in a connected network. Central cliques have applications in corporate, social, communication, power grid, and biological network analysis. Complexity results, bounds, and a combinatorial branch-and-bound algorithm for solving this problem are presented. Computational performance of the proposed algorithm is compared with that of a mixed integer programming technique on a test-bed of randomly generated graphs and real-life networks. 2 - Sparse Model Estimation and its Applications to the Insurance Industry Jessica Wai Yin Leung, University of Sydney, Sydney, Australia, Dmytro Masypura We study the problem of sparse model estimation in the context of binary classification. We formulate the problem by extending the discrete Dantzig selector to accommodate a hinge loss function. To increase the efficiency in solving the problem, we reformulate the problem as a mixed integer linear optimisation problem (MILP) and propose to add additional bounds and warm- start algorithm to obtain near optimal solutions. Preliminary results show that our method is capable of recovering all the true features with high predictive accuracy within a reasonable time frame when the sample size is less than the number of features. 3 - Interdicting Interdependent Smuggling, Money, and Laundering Networks Thomas Sharkey, Rensselaer Polytechnic Institute, CII 5015, RPI, 110 8th St, Troy, NY, 12180, United States, Yeming Shen We consider the problem of disrupting a transnational criminal organization (TCO) operating interdependent smuggling, money, and money laundering networks. The TCO will smuggle contraband across an international border, generate revenue from illegal sales, and then need to send the money back across the border via the laundering network. We examine a bi-level program that seeks to allocate law enforcement resources as to optimally disrupt these interdependent networks. Effective reformulation techniques are discussed as well as examining the impact of disrupting laundering networks across multiple TCOs. 4 - Novel Group Centrality Metrics for Studying Essentiality in Protein Protein Interaction Networks Saeid Rasti, North Dakota State University, Fargo, ND, United States We propose a set of novel group centrality metrics and show their performance in estimating protein importance in protein-protein interaction networks. These centrality metrics are extensions of well-known nodal centrality metrics, such as degree, betweenness, and closeness, for a set of nodes which is required to induce a specific pattern. The structures investigated here range from the stricter induced stars and cliques, to a looser definition of an induced representative structure. We then propose mixed integer programming formulations to solve the problem exactly. Finally, the performance of the proposed metrics in identifying essential proteins in a series of organisms is indicated.

Joint Session DEA/Practice Curated: Applications II Emerging Topic: Productivity, Efficiency and Data Envelopment Analysis Emerging Topic Session Chair: Umit Saglam, East Tennessee State University, Department of Management and Marketing, P.O. Box 70625, Johnson City, TN, 37614, United States 1 - A Dea-based Empirical Analysis for Dynamic Performance of China’s Regional Coke Production Chain Panpan Xia, University of Science and Technology of China, Hefei, China, Jie Wu, Xiang Ji Coke plays a critical role in China’s national economic activities in the past several decades. However, because of the twofold pressures from the sustainability- concerned public and the international steel market downturn, China’s coke industry steps into a dilemma. To help the industry solve its current problems, an empirical analysis for dynamic performance of China’s regional coke production chain is demonstrated. Through adopting the slacks-based measure (SBM) in Data Envelopment Analysis (DEA) and a famous dynamic network DEA framework, this paper simplifies the coke production chain into a three-stage process, and captures the interactions between intermediates inside each stage. 2 - Axiomatic Modeling of Fixed Proportion Technologies Xun Zhou, Doctoral candidate, Aalto University School of Business, Runeberginkatu 22-24, Helsinki, 00100, Finlandi, Timo Kuosmanen Understanding substitution possibilities of inputs/outputs is critical for efficient resource allocation and firm strategy. There are several important examples of fixed proportion technologies where some inputs and/or outputs are not substitutable. However, there is widespread confusion about appropriate modeling of fixed proportion technologies in DEA. We point out and rectify some misconceptions in the published literature, and show how the fixed proportion technologies can be correctly incorporated into the axiomatic framework. 3 - Technical Efficiency in the Chilean Higher Education System: A Comparison with Traditional Measurements of Efficiency Gianfranco Cossani, Universidad Catolica del Norte, Avenida Angamos 0610, Antofagasta, 1270709, Chile, Hernan Caceres, Loreto Codoceo, Jorge Tabilo Data Envelopment Analysis (DEA) has often been used to evaluate the efficiency of higher education institutions in many countries. In Chile, few studies using this technique are available. In this work, we propose a network and dynamic DEA model where each university has a structure with resources shared by teaching and research. For our study case, we examined data from 2013 to 2017 of the Chilean System, and we compared our findings with efficiency scores obtained with four different DEA methods, in addition to rankings commonly used by the government and media outlets. 4 - A Two-stage Performance Assessment of Utility Scale Wind Farms in Texas Umit Saglam, Assistant Professor, East Tennessee State University, Department of Management and Marketing, P.O. Box 70625, Johnson City, TN, 37614, United States A two-stage Data Envelopment Analysis (DEA) models are applied to evaluate productive efficiencies of the 95 large utility-scale wind farms’ electricity generation in Texas, by using pre-determined three input and two output variables. The slack analysis and projection data are obtained for inefficient wind farms to find out benchmarking input-output variables. The sensitivity analysis is provided for the robustness of the DEA models with different combinations of input and output variables of the original model. Tobit regression models are conducted to investigate the reasons for inefficiency. 5 - An Approach for Autonomous Target Selection for an Agent Swarm Barin N. Nag, Professor, Towson University, Department of E-Business & Tech Management, College of Business & Economics, Towson, MD, 21252, United States, Sungchul Hong, Xiaoyin Wang Decision making in target selection for a swarm of drones in military use is a complex process that requires the simultaneous consideration of a number of parameters, some of which may not be known with certainty. An approach is presented here for autonomous targeting decisions. The method uses a combination of Bayesian estimation to overcome problems of uncertainty, and Analogical Reasoning to aggregate a large number of observations to achieve a high level of confidence.

373