2015 Informs Annual Meeting

TA30

INFORMS Philadelphia – 2015

TA29 29-Room 406, Marriott Applications of Analytics I

4 - Distributed Computation of Pareto Sets Margaret Wiecek, Department of Mathematical Sciences, Clemson University, Clemson, SC, 29634, United States of America, wmalgor@clemson.edu, Brian Dandurand The needs of multidisciplinary engineering design have motivated the development of a distributed solution approach to computing Pareto solutions to nonconvex decomposable multiobjective optimization problems. Existing results on augmented Lagrangian coordination techniques and the block coordinate descent method are extended into the multiobjective setting. These convergence analyses lead to a MultiObjective Decomposition Algorithm (MODA) that is applied to packaging in automotive design. TA28 28-Room 405, Marriott Auctions for Ad Space Cluster: Auctions Invited Session Chair: Ian Kash, Microsoft Research, 21 Station Road, Cambridge, United Kingdom, iankash@microsoft.com 1 - General Truthfulness Characterizations via Convex Analysis Rafael Frongillo, Postdoctoral Fellow, Harvard University, 29 Oxford Street, Cambridge, MA, United States of America, raf@cs.berkeley.edu, Ian Kash We present a model of truthful elicitation which generalizes and extends mechanisms, scoring rules, and related settings. Our main result is a generalization of previous characterizations, including a new one for scoring rules on non-convex sets of distributions. We generalize this model to eliciting some property of the agent’s private information, and provide the first general result for this setting. We also show how this yields a new proof of a result in mechanism design due to Saks and Yu. 2 - Inefficiency and Low Revenue in GSP Auctions Indranil Chakraborty, indro@nus.edu.sg In symmetric-Nash equilibrium GSP auction is efficient and generates more than VCG revenue. The revenue ranking may fail under Nash conditions but not when the maximum GSP revenue is considered. We let bidders to have values from conversions instead of clicks, and assume recency effect on conversion to show that in a broad range of situations the GSP auction is inefficient. When efficient equilibria exist the maximum efficient equilibrium revenue in many situations is lower than the VCG revenue. 3 - Optimising Trade-offs Among Stakeholders in Ad Auctions David Kurokawa, PhD Student, Carnegie Mellon University, Computer Science Department, 5000 Forbes Ave, Pittsburgh, PA, 15213, United States of America, dkurokaw@cs.cmu.edu, Yoram Bachrach, Sofia Ceppi, Ian Kash, Peter Key We examine trade-offs among stakeholders in ad auctions. Our metrics are the revenue for the auctioneer, number of clicks for users, and welfare for advertisers. We show how to optimize linear combinations of these utilities via a GSP auction with a per-click reserve price. We then examine constrained optimization of these utilities. Finally, we examine a richer setting that allow using the screen real estate in various ways. 4 - Mechanism Design for Mixed Ads Ian Kash, Microsoft Research, 21 Station Road, Cambridge, United Kingdom, iankash@microsoft.com, Sofia Ceppi, Reza Khani, Yoram Bachrach, Peter Key The GSP auction works when ads are simple, but does not generalize to richer settings. Truthful mechanisms, such as VCG do. However, a straight switch from GSP to VCG incurs significant revenue loss. We introduce a transitional mechanism which mitigates this revenue loss. The mechanism is equivalent to GSP when nobody has updated their bid and is equivalent to VCG when everybody has updated. Our mechanism is based on a technique for making payment functions into a truthful mechanism.

Sponsor: Analytics Sponsored Session Chair: Tarun Mohan Lal, Mayo Clinic, mohanlal.tarun@mayo.edu 1 - Analytics: A Conceptual Framework Robert Rose, President, Optimal Decisions LLC, 4 Kirby Lane, Franklin Park, NJ, 08823, United States of America, optimaldecisions@mac.com The term analytics emerged, and went viral, in November 2005. Since then, there have been many different, and often contradictory, definitions proposed for analytics. Questions such as whether or not analytics is a discipline, and what type of relationship it has to disciplines such as statistics, computer science and operations research, have been unanswered. In this talk, a framework will be presented that explains the emergence of analytics, and logically relates it to other disciplines. 2 - Spreadsheet Software for Linear Regression Analysis Robert Nau, Professor, Duke University, Fuqua School of Business, Durham, NC, 27708, United States of America, robert.nau@duke.edu Spreadsheet add-ins for statistical analysis vary widely in terms of their user interfaces, the detail and design of their output, and support for best practices of analysis. This talk will give a brief overview of some of the market-leading products and compare their linear regression features with a free add-in, RegressIt (http://regressit.com), which was originally designed for teaching an advanced elective on forecasting at Duke University and is now publicly available and widely used. 3 - Strategic Research: Analytics Excellence Charity Maynard, Senior Health System Engineer, Mayo Clinic, 200 First Street SW, Rochester, MN, 55905, United States of America, Maynard.Charity@mayo.edu Mayo Clinic has a long rich legacy of integrating analytics, operations research and industrial engineering in clinical care, research and education for achieving excellence. This presentation will highlight the legacy, sophisticated infrastructure, novel application and dissemination of learning. Key success factors for leveraging analytics and engineering to address the formidable Tarun Mohan Lal, Mayo Clinic, mohanlal.tarun@mayo.edu With the growing trend and concern surrounding health care workforce shortages, there is an increasing call for the redesign of office practices to reduce inefficiency and improve capacity through better use of existing office staff. In this presentation, we will discuss some innovative models of care delivery such as increase pre-visit work, non-face face visits being implemented at Mayo Clinic that has potential for improved operational performance and staff satisfaction. TA30 30-Room 407, Marriott Sports and Entertainment Contributed Session Chair: Stephen Hill Assistant Professor, UNC Wilmington, 601 South College Road, Wilmington, NC, 28403-5611, United States of America, hills@uncw.edu 1 - Does the Number of Days between Professional Sports Games Really Matter? Keith Willoughby, University of Saskatchewan, 25 Campus Drive, Saskatoon, SK, S7N 5A7, Canada, willoughby@edwards.usask.ca, Trevor Hardy In the sport of football, teams typically play at most one game during a week. Leagues may insert “bye weeks” into the schedule for each team, thereby permitting rest and recuperation for players. Teams may feel unfairly treated if they have more (or less) rest than their opponents. We explored over ten years of regular season results from the Canadian Football League to determine if the number of days off experienced by a team between games impacts team performance. challenges in health care today and tomorrow will be discussed. 4 - Analytics to Support Innovation in Outpatient Care Delivery Processes

267