Informs Annual Meeting Phoenix 2018

INFORMS Phoenix – 2018

TD26

3 - A Confidence Interval-based Learning Algorithm for Stochastic Dynamic Programs and its Applications Nan Chen, Chinese University of Hong Kong, William M.W. Mong Engineering Bldg, Rm 609, Shatin N.T, Hong Kong, Xiang Ma We employ the information relaxation technique in this paper to develop a method of value iteration to solve stochastic dynamic programs. The advantages of the new method are that we can construct valid confidence interval to assess the performance of a heuristic policy and provide an recursive algorithm to improve the policy quality. Various applications in finance and inventory management are discussed. Our method can also be easily extended to the investigation on reinforcement learning, in which the decision maker has to learn the dynamic of the stochastic system. 4 - Bitcoin Transaction Fees Wei Jiang, National University of Singapore, Singapore, Min Dai, Steven Kou, Cong Qin Although Bitcoin mining activities are widely discussed in the media, as miners attempt to profit from transaction fees, there are few academic papers studying the transaction fees. We proposed a model for Bitcoin transaction fees from the miners’ perspective. The model is rich enough to incorporate both inventory and demand levels. The model is calibrated to the empirical data and dynamics of the average transaction fees are discussed. The model is made possible by significantly extending the classical Hotelling model for exhaustible natural resources, via the addition of both feedback supply and an S-shaped stochastic demand function. n TD24 North Bldg 131B Practice- Blockchain and Innovation Contributed Session Chair: Qianyu Hu, Penn State University, State College, PA, 16801, United States 1 - Automation and Ecosystem Level Operations Management via Decentralized Ledger Technology Michael Zargham, CEO, BlockScience, Oakland, CA, 94609, United States, Matthew Barlin Incentive functions are engineered to induce independent actors to provide services to decentralized networks. Formal models of the system behavior and careful scoping of models based on design goals is paramount. System models may be derived from game theoretic or behavioral economic models; Operations Research and Swarm Robotics provide existing tools for engineering emergent dynamics. Partitioning the problem improves feasibility and introduces testing requirements and integration challenges for simulation and verification. After deployment, systems require tuning capabilities, monitoring, and maintenance. The safety of public economic infrastructure is a duty of engineers. 2 - Blockchain-based Renewable Energy Certificates Program Fangyuan Zhao, Tsinghua University, Shenzhen, Chinan, Ying Kong, Xin Guo In the era of green energy revolution, distributed renewable energy generation is emerging, while it is out of the scope of conventional renewable energy certificate programs, which involves only large power plants and utility companies. Blockchain shows an advantage of incorporating a galaxy of distributed prosumers in a transparent and low cost manner. This paper proposes a proof of concept model of blockchain-based renewable energy certificates trading system, in which a cryptocurrency is designed for tracing the footprint and stimulating the popularization of renewable energy. The concept model is implemented on Ethereum blockchain for performance and risks analysis. 3 - Blockchain-enabled Pharmaceutical Supply Chain System: A Hyperledger Fabric Design Qianyu Hu, Penn State University, State College, PA, 16801, United States The drug supply chain safety has gained immense attention due to prevalent counterfeit products, widespread fake online pharmacies, drug abuse crisis and so on. Serialization, placing unique identifiers on products, can provide some visibility to the pharma supply chain. However, the identifiers suffer from duplicates and are prone to theft. Our blockchain-enabled system keeps the pharma supply chain data in an immutable and distributed manner to enforce zero counterfeits and allows for transparency and accountability. Data stored on the hyperledger fabric system are accessed based on consented permissions to protect users’ privacy.?

n TD25 North Bldg 131C Complex Service Sponsored: Service Science Sponsored Session

Chair: Fengming Cui, Tsinghua University, Tsinghua, Beijing, China 1 - The Influence of Water, Sanitation and Hygiene Factors on Healthcare Service Utilization During Floods Niratcha Tungtisanont, Assistant Professor, University of Maryland, College Park, 3422 Van Muching Hall, LBPP Dept., College Park, MD, 20742, United States, Aleda Roth We demonstrate that household level factors associated with access to clean water, sanitation and hygiene (WASH) play an integral role on the over- and under-utilization of healthcare resources during a major flood. We apply econometric analyses using field and archival data from 34,000 households in Thailand during the 2011 flood. 2 - The Impact of Dealer Networks on Market Success of Manufacturers Sina Golara, Postdoctoral Associate, Massachusetts Institute of Technology, 1 Amherst St., E40, 239C, Cambridge, MA, 02142, United States, Kevin Dooley Most durable goods manufacturers use a network of dealers to distribute their products. While dealers are the consumer-facing stage of the supply chain and can potentially have a make or break effect on manufacturer’s business, the extent of their influence is not very well understood and has been subject to controversy. By scraping web, we obtain daily sales and consumer rating information for all dealerships in the US and all new cars in inventory in five car classes. We find that consumers’ satisfaction with dealerships influences their purchase of the product. Furthermore, this effect is stronger in markets where the focal brand’s dealer network is sparse and the competing dealer networks are dense. 3 - Performance Effects of Diversity in Experience in Fluid Teams Antti Tenhi l , IE Business School, Calle Maria de Molina 12, piso 5, Madrid, 28006, Spain, Constantin Alba, Fabrizio Salvador How does team diversity in past work experience influence project productivity and responsiveness? We study this question using work assignment data from a multinational software services firm. We operationalize diversity in experience along the dimensions of variety of specializations, separation of experience profiles, and disparity of expertise, hypothesizing and observing distinct curvilinear effects for each dimension. We further theorize that team familiarity may moderate these effects, receiving partial support for these hypotheses. 4 - Agent-based Modeling in Complex Service Systems and Service Systems Engineering Fengming Cui, Tsinghua University, Beijing, China, Lefei Li We introduce agent-based simulation into the design of complex service systems. Each individual in service processes is modeled as a “smart agent” who interacts with each other. By integrating simulation and service science, we obtain insights about service systems dynamics which help us propose better operation strategies. n TD26 North Bldg 132A AI in Service Sponsored: Service Science Sponsored Session Chair: Ming-Hui Huang, National Taiwan University, Taipei, 10617, Taiwan 1 - Advertising Campaign Optimization Niloy J. Mukherjee, Research Scientist, Amazon.com, Seattle, WA, United States Creating an advertising campaign requires several decisions. Including the selection of optimal keywords, match-types and bid values. We discuss methods of making these decisions in an optimal manner when creating a campaign as well as updating them based on performance.

355