Informs Annual Meeting 2017

WA42

INFORMS Houston – 2017

WA41

WA42

352F Decision Analysis Contributed Session Chair: Seyede Yasaman Amirkiaee, University of North Texas, Denton, TX, United States, SeyedeYasaman.Amirkiaee@unt.edu 1 - Inferential Modelling and Decision Making under Uncertainty Jian-Bo Yang, Professor, University of Manchester, F36 MBSE, Booth Street West, Manchester, United Kingdom, jian-bo.yang@manchester.ac.uk, Dong-Ling Xu, Chao Fu, Xiaobin Xu This paper is aimed to investigate issues related to how to make the best use of imperfect data collected from different sources routinely for scientific inference & robust decision making. A data-driven method is introduced & numerical examples are examined to show how evidential reasoning (ER) can be conducted to implement and generalise Bayesian inference in situations where data can be incomplete, ambiguous or inaccurate with different degrees of reliability & weight. A real life case study on the diagnosis of irregularity faults of railway track is conducted to demonstrate a practical procedure for developing an ER-based diagnosis method. 2 - Utilization of Statistical Approaches for Comparing Association Rule Mining Algorithms Sanam Azadiamin, Ohio University, 35 N Mckinley Ave. Apt 112, Athens, OH, 45701, United States, sa349416@ohio.edu, Gulser Koksal Finding the most appropriate association rule mining algorithm according to its application area is studied using statistical approaches. For this purpose, test data is developed based on a statistical experimental design and logistic regression. Interesting rules are found considering pre-specified support level. Finally, the comparison is done compromising over several measures and using the interesting rules. Illustrative examples are provided. 3 - An Adaptive Heuristic for Feature Subset Selection Based on Feature Complementarity Sumanta Singha, Doctoral Student, University of Kansas, 1654 Naismith Drive, Capitol Federal Hall, Lawrence, KS, 66045, United States, sumanta.singha@ku.edu, Prakash P. Shenoy The paper presents a novel approach to feature subset selection, which optimizes relevance, redundancy, and complementarity using an adaptive framework, and determines the trade-off rule endogenously based on these factors. The heuristics explicitly characterizes complementarity, and allows the redundancy- complementarity ratio to guide the search process. Using benchmark datasets, the proposed heuristic outperforms many existing methods of feature subset selection. 4 - Networked Pattern Recognition Framework for Classification Safak Yakti, Research Assistant, Binghamton University, Johnson City, NY, 13790, United States, syakti1@binghamton.edu, Salih Tutun, Mohammad Khasawneh This research aims to propose a comprehensive new framework, namely Networked Pattern Recognition (NEPAR) Framework for different applications. The NEPAR and five different classification methods (SVM, NB, LR, DT, and kNN) are improved by adding information from the proposed network metrics. Information from observations is extracted by building the network, and feature properties for each observation are used to classify the output. Six different datasets (Australian credit approval, diabetes, breast cancer, abalone, SturPlus fMRI, German credit) are used to show the framework outperforms other traditional classification. 5 - Enhancing Lasso’s Sparsity via Mixed Integer Quadratic Programming Ahmed M. Marzouk, Southern Methodist University, Lyle School of Engineering, Dallas, TX, 75275-0123, United States, Marzouk Halit Uster We present a Mixed Integer Quadratic program (MIQP) to find a sparse solution for linear regression (LR) models. We show that the proposed MIQP provides sparser solutions than that provided by LASSO without any degradation in the LR prediction power. Specifically, the average improvement in sparsity when using MIQP was 32.3% and the average computational time was less than one minute. 6 - Public Opinion on Self-driving Vehicles: A Comment Mining Approach Seyede Yasaman Amirkiaee, PhD Student, University of North Texas, 1307 West Highland Street, BLB 375D, Denton, TX, 76201, United States, SeyedeYasaman.Amirkiaee@unt.edu, Nicholas Evangelopoulos Automated driving vehicles are disrupting the transportation industry. As many top motor vehicle and high-tech companies are entering this market, it is important to look at public attitude towards this latest technology. While traditional research has relied on structured surveys, in this research, we are reviewing public opinion by mining posted comments on published news stories, promotional videos, and micro-blogging web sites (Twitter).

360A Planning Contributed Session Chair: Anparasan Mahalingam, Purdue University, West Lafayette, IN, United States, anparasan@purdue.edu 1 - Silence in Identity Salience : the Perks and Downsides of Spanning Multiple Categories on Forming Collaboration Ties Pyung Nahm, Seoul National University, 1 Gwanak-ro, Gwanak-gu, Seoul, 08826, Korea, Republic of, joeynahm@gmail.com We explore the effects having an organizational identity that span multiple categories on the actor’s formation of collaboration ties. We posit that actors without salient identities, spanning multiple categories, are likely to face ‘multiple-category discount’ when selecting potential partners. These actors are more likely to form ties with unexperienced alters because experienced alters question the competence of actors spanning diverse areas. Reputation, a stronger social judgment for competence, will eventually moderate the discounting effect. We utilized co-author network in the management field, analyzing all researches published in the top journals from 2000 to 2016. 2 - When Does Paying More Pay Off? Hazhir Rahmandad, Associate Professor, MIT, 100 Main Steet, E62-442, Cambridge, MA, 02142, United States, hazhir@mit.edu, Zeynep Ton Can well-paying and satisfying jobs accompany, or lead to, profitability in low- cost service sector? Building on qualitative data from a few case studies we map out the systemic feedback mechanisms that could enable good jobs in services. Using a simulation model we show that on a strategy space defined by two dimensions of task richness and compensation, two distinct local profitability peaks emerge: one with low compensation and task richness, representing employee cost minimization, and the good jobs peak with high compensation and task richness. Simulations establish the conditions under which each peak dominates, and the challenges of moving to the good jobs peak due to temporal tradeoffs. 3 - Untangle Interorganizational Collaboration from Value Cocreation Perspective marian.wen@gmail.com, Shih-Chieh Fang, Ching-Ying Huang Received literature on inter-organizational collaborations emphasizes value capture more than how value is jointly created. This research adopts value co- creation approach from service science to address (1) the variety of value that can be created, and (2) key factors making multiple organizations co-create value, then present a tentative research framework. Theoretically, this paper extends inter-organizational value creation by going into the multi-partner settings and promoting multiple organization collaborations. Practically, this paper reflects collaborations with other organizations can be viewed from a collaboration- oriented manner to make pies bigger. 4 - Forming Balance in Status Between Female Directors and Male CEOs: Through Adjusting Male CEO’s Economic Status Junghyun Mah, Seoul National University, KangnamGu, Irwonbondong Mokryun Town 107-601, Seoul, 06355, Korea, Republic of, junghyunmah@gmail.com Little has been explored on how the subdivision of status for female directors and male CEOs invoke social comparison when determining male CEOs’ compensation. Female directors enjoy the high economic status, whereas they occupy low social status compared to male CEOs. In contrast, male CEOs occupy high economic and social status. Status inconsistent female directors aim to balance their overall status with status consistent male CEOs through targeting male CEOs’ economic status. It is observed by decreased growth rate of male CEOs’ compensation, narrowed pay disparity, increased shareholder’s wealth, and the number of board appointments by female directors moderates the influence. 5 - Information Systems Integration in Mergers and Acquisitions Aditya N.Saharia, Associate Professor, Fordham University, Gabelli School of Business, 140 West 62nd Street, New York, NY, 10023, United States, saharia@fordham.edu Often corporate mergers fail because (or create negative values) because of the difficulty in integrating IT systems and the operational and management processes they support. In this work we use a knowledge based view of the firm to identify critical success factors for post-merger IT integration. Marian Dan-Wei Wen, National Cheng Kung University, No. 1, University Rd., East Dist.,, Tainan, 70101, Taiwan,

455