2016 INFORMS Annual Meeting Program

MB42

INFORMS Nashville – 2016

MB42 207D-MCC Crowd-Commerce Applications in Operations and Revenue Management Sponsored: Revenue Management & Pricing Sponsored Session Chair: Rene A Caldentey, University of Chicago, Chicago, IL, United States, rene.caldentey@chicagobooth.edu Co-Chair: Yifan Feng, The University of Chicago, 5807 S Woodlawn Ave, Chicago, IL, 60637, United States, yfeng4@chicagobooth.edu 1 - Learning Customer Preferences Through Crowdvoting Yifan Feng, University of Chicago, Chicago, IL, 60637, United States, yfeng4@chicagobooth.edu, Rene A Caldentey, Christopher Ryan We study a seller introducing a new product with multiple potential product designs into the marketplace. In order to pick the design that is most likely to be preferred by customers, the seller uses an online system that allows potential buyers to vote for their preferred designs. We study how to dynamically customize each individual voter’s choice set, in order to most efficiently learn overall customer preferences. We propose an algorithm that balances breadth of choice and accuracy in determining the best product. We show this algorithm is asymptotically optimal in speed of learning. 2 - Simultaneous Vs. Sequential Crowdsourcing Contests Lu Wang, University of Toronto, Rotman School of Management, 105 St George Street, Toronto, ON, M5S 3E6, Canada, lu.wang12@rotman.utoronto.ca, Ming Hu In a crowdsourcing contest, innovation is outsourced to an open crowd. We consider two alternative crowdsourcing mechanisms for an innovative product involving multiple attributes. One is to run a simultaneous contest, where the best is selected from the single solution simultaneously submitted by each contestant. The other is to run multiple sequential sub-contests, with each dedicated to one attribute and a later sub-contest built on the best outcome from earlier sub-contests. While both mechanisms have their own advantages, either could win over depending on situations. 3 - Contests And Inequality Mohamed Mostagir, University of Michigan, Ann Arbor, MI, United States, mosta@umich.edu, Yesim Orhun, Hamidreza Tavafoghi Contests are one of the standard mechanisms that firms employ to extract the most effort from participants, whether these participants are crowd workers or the firm’s own personnel. We study contests that are repeatedly played by the same agents, with a focus on how information revelation about past play impacts future efforts. We show that such revelation can be detrimental to aggregate effort, and discuss how regulations (e.g. the SEC Dodd-Frank act) that require employers to reveal wages in an attempt to curb inequality can lead to unexpected effects that ultimately result in higher inequality amongst workers in environments that resemble contests, i.e. where wages follow a rank-based structure. MB43 208A-MCC Applied Decision Analysis Sponsored: Decision Analysis Sponsored Session Chair: Saurabh Bansal, Penn State University, Penn State University, State College, PA, 16802, United States, sub32@psu.edu 1 - Analyzing Both The Cost And Strategic Value Of Sustainable Supply Chains The Brazilian government will require the use of additives in all gasoline fuel starting in 2017. We use optimization modeling to help our industry partner design their new supply chain network and study the cost of reducing carbon emissions. However, we then use decision analysis to study the strategic value of sustainable supply chain designs in obtaining market share. In our case study, the strategic value outweighs the cost of reducing emissions. Jason Merrick, Virginia Commonwealth University, jrmerric@vcu.edu, Paul Brooks, Lance Saunders

2 - Eliciting Newsvendor Quantile: Direct Or Decomposed Assessments? Saurabh Bansal, Penn State, sub32@psu.edu

We consider the newsvendor problem that is commonly used in practice. We report the results of a laboratory study in which participants provide (i) direct solution to the problem, (ii) decomposed solution to the problem. Our results help identify the optimal discretion levels that should be provided to managers. 3 - Supporting The Prioritization Of Emerging Animal Health Threats For The UK Department Of Agriculture With Decision Analysis Gilberto Montibeller, Loughborough University, Loughborough, United Kingdom, g.montibeller@lboro.ac.uk Gilberto Montibeller, Decision Consulting Ltd., Leicester, United Kingdom, g.montibeller@lboro.ac.uk, L. Alberto Franco Emerging animal health threats pose serious risk to humans and countries, and represent a serious challenge for both analysts and policy makers. We employed a decision analytic framework to develop a risk management support system to help the UK Department of Agriculture (DEFRA) with the prioritisation of such threats, providing an effective mechanism for ranking them and supporting the design of policy recommendations. The system is supporting the recommendations of DEFRA’s Veterinary Risk Management group since 2009. Benefits for the client include increased rigour in evidence gathering, transparent assessments, and a traceable and more streamlined decision process. 4 - How Did We Integrate Optimization And Machine Learning In our Solution Tool at Mckinsey Halil I Cobuloglu, Sr. Research Analyst, McKinsey & Company, 404 Wyman Street, Waltham, MA, 02451, United States, halil.cobuloglu@gmail.com, Dimitris Bertsimas, Nathan Uhlenbrock, Prodipto Ghosh In this project, we have developed a territory optimization tool for our clients. In order to reach solution fast, we have integrated various algorithms including machine learning and optimization techniques in our model. This tool helps companies efficiently use their limited sources and optimize their territories with more balanced workload. MB44 208B-MCC Panel: Advice from Award Winning Researchers Sponsored: Decision Analysis Sponsored Session Moderator: Andrea Hupman Cadenbach, University of Missouri - St. Louis, St Louis, MO, United States, cadenbach@umsl.edu This panel discussion features several distinguished researchers who have won awards from the Decision Analysis Society. Panelists will discuss the processes behind their research that contributed to their success and share advice for junior faculty, postdoctoral researchers, and PhD candidates. 1 - Panelist L Robin Keller, Professor, University of California - Irvine, Irvine, CA, United States, LRKeller@uci.edu 2 - Panelist James S Dyer, University of Texas - Austin, j.dyer@mccombs.utexas.edu 3 - Panelist Robert Clemen, Duke University, clemen@duke.edu 4 - Panelist Ali E Abbas, Professor and Director of DECIDE, University of Southern California, Los Angeles, CA, 90089, United States, aliabbas@usc.edu

162