Informs Annual Meeting Phoenix 2018

INFORMS Phoenix – 2018

SC24

2 - Adaptive Decision Models for Smart Service Systems Ralph D. Badinelli, Virginia Polytechnic Institute, Dept of Business Information Tech (0235), Blacksburg, VA, 24061, United States We describe an adaptive decision-modeling framework for integrating IoT and BDA into decision support for smart service systems. The sensing capabilities of IoT and the predictive capabilities of BDA are often mistaken for smartness. However, the integration of these analytic technologies into adaptive decision models is necessary to create a smart DSS. In particular, the engagement decisions of actors in a service ecosystem are the locus of smartness in a service. These decisions are dynamic, stochastic and of high dimensionality. Approximate dynamic programming and heuristics are necessary.

n SC24 North Bldg 131B Prosocial Behavior on Digital Platforms Sponsored: EBusiness Sponsored Session Chair: Xue Tan, Indiana University, Bloomington, IN, 47405, United States 1 - Micro-giving: On the Use of Mobile Devices and Monetary Subsidies in Charitable Giving Dokyun Lee, Carnegie Mellon University, Tepper School of Business, 5000 Forbes Avenue, Pittsburgh, PA, 15213, United States, Dongwon Lee, Anandasivam Gopal This study examines the effects of push notifications, monetary subsidies, and intertemporal choices of rewards on donation outcomes by individual users. Our experiments provide three main findings. First, push notifications have a remarkably high effect on donation behavior in terms of the decision to donate as well as donation amounts, especially for idle users. Second, offering rebates as economic incentives are more effective than offering matching grants, in particular for male consumers. Finally, while users are indifferent between receiving their rewards now versus later, users that are offered a rebate respond Lusi Yang, National University of Singapore, 15 Computing Drive, COM2-01-02, Singapore, 117418, Singapore, Jungpil Hahn, David Lehman This study aims to uncover the dynamics of crowdfunding by examining how fundraising patterns affect subsequent entrepreneurial strategies in managing the project during the funding process. Toward this goal, drawing insights from the behavioral theory of the firm, we identify two types of entrepreneurial strategies from potential courses of actions, i.e., exploitation and exploration, and examine the influence of concurrent fundraising performance on the two types of strategic actions. 3 - You Run, I Donate: The Impact of Corporate Sponsorship in a Three-sided Charitable Platform Xue Tan, Indiana University, Kelley School of Business, 1309 E. 10th St, HH 4131, Bloomington, IN, 47405, United States, Lu Yan, Alfonso J. Pedraza-Martinez This paper seeks to understand a three-sided marketplace where corporate sponsors individuals’ charitable efforts by donating to charities based on the individuals’ running miles captured by an mobile application. Using social media data that captured a natural experiment, we reveal the impact of announcing and showing gratitude to corporate sponsors. 4 - Influence of Peer Based Monetary Incentive on Content Generation Zixuan Meng, University of Washington, Seattle, WA, 98115, United States, Yi-Chun Ho, Junjie Wu, Yong Tan User-generated content has been widely studied in the past couple of years. Platforms have tried multiple methods to encourage the content generating behavior, and some have started motivating through monetary incentives. Using a unique setting where content consumers can voluntarily support writers through direct money transaction, we analyze how monetary incentive influences writers behavior. Our findings show that the availability of the monetary incentive encourages writers to increase both the quality and the quantity of writing. However, if the monetary incentive is below writers’ expectation, writers can be demotivated. n SC25 North Bldg 131C Analytic Support for SMART Service Systems Sponsored: Service Science Sponsored Session Chair: Ralph D. Badinelli, Virginia Polytechnic Institute, Blacksburg, with greater alacrity when rewarded immediately rather than later. 2 - Fundraising Patterns and Entrepreneurial Strategies in Crowdfunding

n SC26 North Bldg 132A

Undergraduate Operations Research Prize I Emerging Topic: Undergraduate Operations Research Prize Emerging Topic Session Chair: Jennifer A. Pazour, Rensselaer Polytechnic Institute, 110 8th Street, CII 5217, Troy, NY, 12180, United States 1 - Numerical Analysis of the Knowledge Gradient for Locally Quadratic Functions in Multidimensional Learning Settings Michael Li, Princeton University, Princeton, NJ, United States, Warren Powell This paper studies the knowledge gradient for locally quadratic functions (KGLQ), a new policy for optimizing smooth, expensive functions. By capturing structural bias through a tunable parameter, KGLQ encourages more strategic behavior than previous policies. KGLQ is applied to parametric functions, newsvendor functions, and energy arbitrage. Numerical results show KGLQ outperforms its predecessors. However, the optimal tunable parameter values are problem dependent and vary across dimensions. 2 - Centers for Disease Control and Prevention as a Strategic Agent in the Pediatric Vaccine Market: An Analytical Approach Kayla Cummings, Massachusetts Institute of Technology, Cambridge, MA, United States, Banafsheh Behzad, Susan Martonosi The CDC’s significant patronage of pediatric vaccines affords them leverage in preventing monopolies via public sector price negotiation. Our optimization model ensures adequate vaccination levels and manufacturers’ success at minimum government cost. We find that dissimilar products advantageously segment markets with asymmetric manufacturers. Furthermore, markets are at lower risk when high-capacity manufacturers have low R&D costs, especially in cases of high demand and asymmetry. A shift from annual contracts to ongoing negotiations would give the CDC more control over market viability. 3 - Multi-Period Lot Sizing with Supplier Selection: Structural Results, Complexity and Algorithms Meichun Lin, Shanghai Jiao Tong University, Shanghai, China, Woonghee Tim Huh, Meichun Lin We consider a multi-period lot-sizing problem with multiple products and suppliers. We aim to determine the order quantity to minimize the total cost of product purchase and inventory holding under deterministic time-varying demand. We show the intractability of the general problem, analyze structural properties for three polynomial-time solvable cases and propose dynamic programming algorithms. We also develop a heuristic for the general problem and computationally show that it works well. Moving Online: How Exactly to Repackage a Face to Face Class into an Awesome Online Experience Sponsored: Education (INFORMED) Sponsored Session Chair: Carrie Beam, University of Arkansas, Walnut Creek, CA, 94596, United States 1 - Moving Online: How Exactly to Repackage a Face to Face Class into an Awesome Online Experience Carrie Beam, University of Arkansas, Walnut Creek, CA, 94596, United States “I’ve taught in the classroom for 30 years, and now my dean is pressuring me to move my class online. Where to start?” “I’m the dean they are all complaining about. What management techniques do we need so our faculty are able to happily move online?” Here’s a practical roadmap for moving face-to-face classes online. Live examples of online classes in Operations Management, Analytics, n SC27 North Bldg 132B

VA, 24061, United States 1 - On Bias in AI Systems Rama Akkiraju, IBM, San Jose, CA, United States

Some argue that one must strive to build unbiased AI models by training AI models with comprehensive training datasets. In this talk, I’d argue that there is no such a thing as an æunbiased’ Machine Learning model. Instead of striving for unbiased machine learning models, a machine learning model must state its biases openly. Minimizing biases has to start with creating test datasets rather than with training datasets. I will provide a practitioner’s point-of-view on mitigating biases in AI models.

70