Informs Annual Meeting 2017

MB22

INFORMS Houston – 2017

2 - Outcome-driven Personalized Treatment Design for Managing Diabetes Xin Wei, Georgia Institute of Technology, Atlanta, GA, United States, xwei36@gatech.edu, Eva Lee

4 - Markov Chain Models for Controlling the Exposure Frequency Distribution of Online Advertising Ali Hojjat, University of New Hampshire, Durham, NH, United States, ali.hojjat@unh.edu, John G.Turner Recent trends in online advertising show that explicit reach and frequency specifications are preferred over aggregate impression or budget goals. We propose two Markov chain models for serving ads that can achieve a desired frequency distribution for an online ad campaign. We characterize the necessary and sufficient conditions on the arrival distribution that must hold for a feasible ad serving policy to exist, and characterize the publisher’s impression assignment rule in closed form. Extensions of our framework to multi-advertiser and multi- audience-segment settings are also presented. 342E Pricing, Information Sharing, and Other Decisions in RM Applications Sponsored: Revenue Management & Pricing Sponsored Session Chair: Ramandeep Randhawa, University of Southern California, Los Angeles, CA, 90089, United States, ramandeep.randhawa@marshall.usc.edu 1 - Markdown Policies for Demand Learning and Strategic Customer Behavior N. Bora Keskin, Duke University, Fuqua School of Business, 100 Fuqua Drive, Durham, NC, 27708-0120, United States, bora.keskin@duke.edu, Hongfan Chen Consider a firm selling a product to a finite population of myopic and strategic customers with heterogeneous valuations. The firm faces uncertainty about the market size and how strategic customers are. We develop guidelines for designing asymptotically optimal markdown policies for demand learning in the presence of strategic customers. 2 - Contingent Stimulus in Crowdfunding Longyuan Du, Rotman School of Management, University of Toronto, 105 St George St, Toronto, ON, M5S.3E6, Canada, Longyuan.Du14@Rotman.Utoronto.ca, Ming Hu, Jiahua Wu We consider a model where backers arrive sequentially at a crowdfunding project. Upon arrival, a backer makes her pledging decision by taking into account the expected project success rate. We characterize the dynamics of a project’s pledging process. To boost success, we propose and characterize three contingent stimulus policies, i.e., seeding, feature upgrade and limited-time offer. We show that the benefit of contingent policies is greatest in the middle of crowdfunding campaigns. Testing with the data set of Kickstarter, we obtain empirical evidence that the projects’ success rates improve by 14.6% on average with updates in the middle of the campaign and when the pledging progress is lagging. 3 - Bayesian Persuasion on Networks Kimon Drakopoulos, University of Southern California, drakopou@marshall.usc.edu, Ozan Candogan In the presence of misinformation, platforms face a tradeoff between customer engagement and quality of shared information. In this paper we explore this tradeoff and discover the optimal information revelation mechanism. The amount of information provided to each agent according to this mechanism turns out to relate to a novel centrality measure that we further analyze and relate to existing notions of centrality. 4 - Information Sale in Competitive Markets Kostas Bimpikis, Stanford University, kostasb@stanford.edu, Alireza Tahbaz-Salehi, Davide Crapis We study the strategic interaction between a seller of information and buyers that compete in a downstream market. Our analysis illustrates that the nature and intensity of competition are key in determining the optimal strategy. In particular, we show that when the customers’ action are strategic complements, it is to optimal to offer the most accurate information to all customers. In contrast, when actions are strategic substitutes, it is optimal to either (i) restrict the overall supply of the information product, or (ii) offer a product of inferior quality. We also show that the incentive to restrict the supply or quality of information intensifies in the presence of information leakage. MB23

This work is joint with Grady Memorial Hospital. The management of gestational diabetes mellitus(GDM) focuses on close monitoring of a patient’s blood glucose level while the clinician experiments with dosing strategy based on some clinical guidelines and his/her own experience. We propose a drug-effect-based personalized approach to improve treatment outcome for GDM patients. Specifically, a predictive model based on pharmacokinetic and pharmacodynamic(PK/PD) analysis of drug dosage and blood glucose level is proposed to uncover personalized treatment effect and then incorporated in a planning model for optimal dosing strategy. Results for GDM patients will be presented. 3 - A Supervised Biclustering Approach for Personalized Medicine Traditional medicine typically applies one-size-fits-all treatment for the entire patient population whereas precision medicine develops tailored treatment schemes for different patient subgroups. The fact that some factors may be more significant for a specific patient subgroup motivates researchers to develop new approaches to subgroup analysis, which is an effective strategy to personalize treatment. In this study, we propose a novel patient subgroup detection method, called Supervised Biclustring (SUBIC) using convex optimization and apply our approach to detect patient subgroups in a vulnerable demographic subgroup at the high risk of cardiovascular disease. 342D Mechanism Design, Networks, and New Markets II Sponsored: Revenue Management & Pricing Sponsored Session Chair: Santiago Balseiro, srb43@duke.edu Co-Chair: Ozan Candogan, University of Chicago, Chicago, IL, 27708, United States, ozan.candogan@chicagobooth.edu 1 - Strategic Dynamic Pricing with Network Externalities Azarakhsh Malekian, University of Toronto, Toronto, ON, Canada, azarakhsh.malekian@rotman.utoronto.ca We study the optimal pricing of a monopolist selling goods in a dynamic pricing game with multiple rounds. Customers are forward-looking and experience a (positive) network externality. The seller commits to a price sequence and strategic buyers play a perfect Bayesian equilibrium. We fully characterize the optimal pricing policy and show that it is linearly increasing in time, where the slope of the price path is described by a single network measure: sum of the entries of the inverse of network externality matrix, termed network effect. We also study the effect of price discrimination and establish that in the earlier rounds the price offered to more central buyers is lower. 2 - Facilitating the Search for Partners on Matching Platforms: Restricting Agent Actions Yash Kanoria, Columbia Business School, New York, NY, 10027, United States, ykanoria@columbia.edu, Daniela Saban Two-sided matching platforms, such as those for labor, accommodation, dating, and taxi hailing, control many aspects of the search for partners. We consider a dynamic model of search by strategic agents with costly discovery of pair-specific match value. We find that in many settings, the platform can mitigate wasteful competition in partner search via restricting what agents can see/do. If agents on one side have a tendency to be more selective (due to smaller screening costs or greater market power), the platform should disallow agents on the less selective side from reaching out. This forces agents on the more selective side to reach out and improves the welfare of agents on the less selective side. 3 - Temporal Microstructure of Surge Pricing Bin Hu, UNC Chapel Hill, McColl Building CB3490, Kenan-Flagler Business School, Chapel Hill, NC, 27599, United States, bin_hu@unc.edu, Ming Hu, Han Zhu Ride-sharing platforms such as Uber and Lyft contingently increase prices (known as Surge Pricing) to modulate surges of rider requests and attract drivers. However, in managing unpredictable demand shocks with Surge Pricing, an important feature neglected in the existing literature is that supply (drivers) reacts much slower than demand (riders) to price changes. We model this feature and the resulting rider and driver strategic behaviors, and show that neglecting this feature can lead to significantly distorted and underperforming decisions. Milad Zafar Nezhad, Wayne State University, Detroit, MI, United States, m.zafarnezhad@wayne.edu, Kai Yang MB22

172

Made with FlippingBook flipbook maker