2015 Informs Annual Meeting

SB78

INFORMS Philadelphia – 2015

SB76 76-Room 204C, CC Efficient Learning in Stochastic Optimization Sponsor: Simulation Sponsored Session Chair: Ilya Ryzhov, University of Maryland, 4322 Van Munching Hall, Robert H. Smith School of Business, College Park, MD, 20742, Qihang Lin, The University of Iowa, 21 East Market Street, Iowa City, IA, 52245, United States of America, qihang-lin@uiowa.edu, Xi Chen, Kevin Jiao Crowdsourcing is often used as a tool to rank a list of items based on pairwise comparisons. However, the comparison results by crowdsourcing have a low quality due to unreliable workers. To reduce the cost and increase the accuracy of ranking, we propose a multistage strategy where pairs of items are assigned to workers based on a joint learning of item’s ranking and worker’s reliability. The performances of our strategies are evaluated over simulated and real data. 2 - Sequential Allocation for Customer Acquisition Eric Schwartz, University of Michigan, ericmsch@umich.edu, Katie Yang, Peter Fader To acquire customers, firms allocate resources across a range of sources of acquisition, but they are uncertain about which ones are best. Over time they learn about their customers and sequentially reallocate their resources to earn a better return on acquisition spend. We frame the sequential acquisition decisions as a multi-armed bandit problem, and comparing a set of acquisition policies to assess their ability to acquire from the right sources of customers. 3 - Optimal Dynamic Pricing with Demand Model Uncertainty: A Squared-Coefficient-of-Variation Rule Bora Keskin, Duke University, Fuqua School of Business, We consider a price-setting firm that sells a product over a continuous time horizon. The firm is uncertain about the sensitivity of demand to price changes, and updates its prior belief on an unobservable sensitivity parameter by observing demand responses. We derive and solve a partial differential equation to show how the value of learning should be projected onto prices in an optimal fashion. 4 - Expected Improvement is Equivalent to OCBA Ilya Ryzhov, University of Maryland, 4322 Van Munching Hall, Robert H. Smith School of Business, College Park, MD, 20742, United States of America, iryzhov@rhsmith.umd.edu We consider ranking and selection with independent normal observations, and analyze the asymptotic sampling rates of expected improvement (EI) methods in this setting. EI often performs well in practice, but general rate results have been largely unavailable. We prove that variants of EI produce simulation allocations that are essentially identical to those chosen by the optimal computing budget allocation (OCBA) methodology. This is the first general equivalence result between EI and OCBA. SB77 77-Room 300, CC Supply Chain Management II Contributed Session Chair: Ehsan Bolandifar, Assistant Professor, The Chinese University of HongKong, Room 922, 9/F, Cheng Yu Tung Building, No.12, Chak Cheung Street, Shatin, N.T., HongKong, Hong Kong - PRC, ehsan@baf.cuhk.edu.hk 1 - Cooperative Replenishment in the Presence of Intermediaries Behzad Hezarkhani, Assistant Professor, Nottingham University Business School, Jubilee Campus, Nottingham, United Kingdom, behzad.hezarkhani@nottingham.ac.uk, Marco Slikker, Tom Van Woensel In complex supply chains, individual downstream buyers would often rather replenish from intermediaries than directly from manufacturers. Direct replenishment from manufacturers can be a less costly alternative when carried out by the buyers cooperatively. This talk presents a framework for cooperative/non-cooperative replenishment in multi-product supply chains with intermediaries. 100 Fuqua Drive, Durham, NC, 27708-0120, United States of America, bora.keskin@duke.edu United States of America, iryzhov@rhsmith.umd.edu 1 - Cost-efficient Learning for Crowdsourced Ranking

2 - Two-class Single-period Inventory Allocation Policies in Smart Meter Installation Projects Behzad Samii, Vlerick Business School, Ave de Boulevard 21, Brussels, Belgium, behzad.samii@vlerick.com Smart meter device are the costliest elements of rollout projects. Complexity stems from supply inflexibility due to strict tendering procedures and high holding cost due to fast obsolescence. If some partial information regarding the bottom line impact of a shortage in one customer class compared to the other can be conjectured, then we can derive closed form expressions for the expected number of units short in each demand class under commonly used SN and TN nesting allocation mechanisms. 3 - Architecting Fail-safe Supply Networks Shabnam Rezapour, The University of Oklahoma, 2248 Houston Ave., apt # 2, Norman, OK, 73071, United States of America, shabnam_rezapoor@yahoo.com, Amirhossein Khosrojerdi, Janet K. Allen, Farrokh Mistree A fail-safe network is one which mitigates the impact of disruptions and provides an acceptable service level. This is achieved by designing its topology (structurally fail-safe) and coordinating flow dynamics (operationally fail-safe). We analyze the importance of being robust, resilient, and controllable in having structurally fail- safe against disruptions. We show to have an operationally fail-safe supply network, flow dynamics should be reliable against demand and supply-side variations. 4 - Waveless Warehousing ? Why and Why Not ? Adrian Kumar, Exel Inc, 570 Polaris Parkway, Westerville, OH, United States of America, adrian.kumar@exel.com, Manjeet Singh E-fulfillment operations constantly struggle with processing peak volumes quickly due to system, labor and equipment constraints. Waveless is a dynamic order fulfillment method that pulls demand into a resource/sub-system when it becomes available. The dynamic order set is built on optimal real time decisions based on productivity, equipment utilization, etc. This study defines complete and partial waveless systems and discusses the pros and cons of implementing them. 5 - Component Procurement through Group Purchasing Organizations Ehsan Bolandifar, Assistant Professor, The Chinese University of Hong Kong, Room 922, 9/F, Cheng Yu Tung Building, No.12, Chak Cheung Street, Shatin, N.T., Hong Kong - PRC, ehsan@baf.cuhk.edu.hk, Mojtaba Soleimani This paper studies component procurement in a supply chain setting where two competing Original Equipment Manufacturers (OEMs) source a common component from a competitive supply market. We assume that ordering happens after procurement negotiations, i.e., OEMs first compete in the market before they negotiate for their component procurement potentially through Group Purchasing Organizations (GPOs). We show that procurement through a GPO may hurt an OEM with a lower bargaining power.

SB78 78-Room 301, CC Supply Chain Practice and Empirics Contributed Session

Chair: Faraz Ramtin, University of Central Florida, 2011 Puritan Rd, Orlando, FL, 32807, United States of America, faraz.ramtin@ucf.edu 1 - Is Supply Chain Success Emulatable? A Framework of Analogous Learning from Supply Chains and a Case Study

Violette Wen, PhD Student, The University of Auckland, 12 Grafton Rd, CBD, Auckland, 1010, New Zealand, violette.wen@auckland.ac.nz, Tiru Arthanari

We explore the feasibility of emulating from one successful supply chain to another produce line in agri-fresh produce. The case will study the state-of-art New Zealand kiwifruit supply chain and provide a framework with key enablers and disenablers for transferring knowledge to the struggling Xinjiang Hami melon industry in China. The research will provide rich empirical evidence about a developing country’s agri-fresh supply chain at various levels. 2 - The Inventory Value of Cross Docking in a Supply Chain: An Empirical Study Xingyue Zhang, Lehigh University, 621 Taylor Street, Bethlehem, PA, 18015, United States of America, xiz313@lehigh.edu, Oliver Yao, Jiazhen Huo, Yongrui Duan Cross docking is a supply chain strategy to enhance supply chain performance by directly moving inbound orders to outbound shipments without storage. Using a large-scale, SKU level data set collected from a large retail chain, we find that cross docking reduces store-level inventory by 101 units on average and that cross docking is more beneficial to reduce inventories for products with higher prices or higher demand rate, or for the stores that are closer to their distribution center.

91