Informs Annual Meeting Phoenix 2018

INFORMS Phoenix – 2018

MD16

2 - Issue Resolution Estimation for Customer Service Centers Han Ye, U. of Illinois at Urbana-Champaign, 350 Wohlers Hall, 1206 South Sixth Street, Champaign, IL, 61820, United States Issue resolution estimation is critical for service center management. Traditionally, it is estimated via surveys and monitoring. In this paper, new models are developed to estimate issue resolution from operational data. The proposed models are then compared to existing practices under various scenarios. Performance grids of each studied model are constructed to provide practical guidelines on model selection. 3 - Can Customer Arrival Rates Be Modelled by Sine Waves? Donald Lee, Yale University, New Haven, CT, 06520, United States, Ningyuan Chen, Haipeng Shen Customer arrival patterns typically exhibit strong seasonal effects. It is therefore natural to ask: Can a nonhomogeneous Poisson process (NHPP) with a rate that is the sum of sinusoids provide an adequate description of reality? We empirically investigate this question in two settings of interest to operations scholars: Arrivals to an emergency department and to a call centre. We develop novel estimation and testing procedures to show that the model is consistent with arrivals data from both settings. Our findings, combined with the flexibility and tractability of sinusoids, suggest that the NHPP with a sinusoidal rate function is a worthy workhorse model for time-varying arrival processes. Socially Sustainable and Inclusive Operations Sponsored: Manufacturing & Service Oper Mgmt/Sustainable Operations Sponsored Session Chair: Beril L. Toktay, Georgia Institute of Technology, Atlanta, GA, 30308, United States 1 - Stable Schedules: A Win for Both Retail Associates and Stores Saravanan Kesavan, University of North Carolina-Chapel Hill, Kenan-Flagler Business School, Cb 3490 Mccoll Building, Chapel Hill, NC, 27599-3490, United States In this study at the Gap Inc., we run a large field experiment in 30 of their stores over a 10 month period to investigate the impact of improving schedule stability of 3000 part-time associates. We follow a randomized encouragement design to allow store managers to implement five interventions in the treatment stores. We document the increase in stability and its impact on sales, productivity, and turnover of employees. 2 - Preventing the Cost of Over Development in China Kwan Yu (Chris) Lo, The Hong Kong Polytechnic University, Hong Kong, Hong Kong, Yi Zhou, Christopher S. Tang The low-cost sourcing strategy of international brands is one of the reasons of over-development in developing countries; the manufacturing firms in developing countries jeopardize society assets for the firms’ economic benefits. We challenge if such short-term economic success could last in today’s transparent global supply chain. Based on the environmental violations data of the listed manufacturing firms published by Institute of Public and Environmental Affair (IPE), we estimate the long-term impact of environmental violations, and identify the key predictors of environmental violations in China. We developed prediction models to identify high-risk firms before they pollute. 3 - Optimal Seeding Policy under Rainfall Uncertainty Ying Zhang, Clemson University, Clemson, SC, United States, Jayashankar M. Swaminathan Increased agricultural productivity is often cited as a solution to the impending global food shortage problem. In this paper, we develop a model to determine the optimal seeding policy under rainfall uncertainty using a finite-horizon stochastic dynamic program. In our model, a farmer needs to decide whether to plant a seed in each period given the soil moisture. We show that the optimal policy is a time- dependent threshold-type policy where the farmer should plant when the seed amount on hand is above the optimal threshold. Utilizing field weather data from Southern Africa, we show significant yield advantage of the optimal schedule over commonly used heuristics. 4 - Effects of Incentive Programs for Workplace Safety Ravi Subramanian, Georgia Institute of Technology, 800 West Peachtree Street Nw, Scheller College of Business, Atlanta, GA, 30308, United States, Vinay Ramani Using analytical models, we contrast the effects of contextual parameters on the occurrence and reporting of safety incidents under outcome-based and behavior- based incentive programs for workplace safety. n MD16 North Bldg 127B

n MD14 North Bldg 126C Joint Session MSOM/Practice Curated: Learning and Information Theory Applications in Queues Sponsored: Manufacturing & Service Oper Mgmt/Service Operations Sponsored Session Chair: Nur Sunar, UNC, UNC, Chapel Hill, NC, 27517, United States 1 - Reinforcement with Fading Memories Kuang Xu, Stanford Graduate School of Business, 655 Knight Way, Stanford, CA, United States We study the effect of imperfect memory on decision making in the context of a stochastic sequential action-reward problem. An agent chooses a sequence of actions which generate discrete rewards at different rates. She is allowed to make new choices at rate , while past rewards disappear from her memory at rate . We provide closed-form formulae for the agent’s steady-state choice distribution in the regime where the memory span is large ( ? 0), and show that the agent’s success critically depends on how quickly she updates her choices relative to the speed of memory decay. 2 - Signaling in Queues with Risk Averse Customers Krishnamurthy Iyer, Cornell University, Ithaca, NY, 14850, United States, David Lingenbrink We study revenue-optimal signaling in an unobservable queue offering service at a fixed-price to a Poisson arrival of customers, who decide to join or balk upon arrival. We focus on the setting where customers are strategic and risk-averse: a customer joins only if the sum of the mean of her waiting time and a multiple of its standard deviation is below a given threshold. Although the revelation principle no longer holds, a restricted form of the principle allows us to formulate an iterative approach to solve the information design problem, where each iteration involves optimizing a linear objective under quadratic constraints. 3 - A Semi-parametric Bayesian Model for Call Center Arrivals Kaan Kuzu, Univ of Wisconsin-Milwaukee, Sheldon B. Lubar School of Business, P.O. Box 742, Milwaukee, WI, 53201-0742, United States, Refik Soyer We describe and analyze data for arrivals to a call center by presenting a modulated Poisson process model, which takes into account both covariate and time effects on the call volume intensity. We introduce a semi-parametric model and develop its Bayesian analysis to assess the effectiveness of different advertising strategies as well as to predict call arrival patterns. The proposed model and the methodology are implemented using real call center arrival data. We show that the proposed semi-parametric model has higher prediction accuracy than prior parametric models in literature. 4 - Dynamic Learning and Rational Customers in Services Nur Sunar, UNC, 1604 Village Crossing Drive, Chapel Hill, NC, 27517, United States, Yichen Tu, Serhan Ziya We study a queueing system where customers can dynamically learn about a service feature. Our analysis shows that such rational customers can help the service provider boost its expected profit. n MD15 North Bldg 127A Joint Session MSOM/Practice Curated: Business Analytics Sponsored: Manufacturing & Service Oper Mgmt/Service Operations Sponsored Session Chair: Han Ye, U. of Illinois at Urbana-Champaign, Champaign, IL, 61820, United States Co-Chair: Haipeng Shen, Hong Kong 1 - On the Accuracy of the Last-to-enter-service Announcement: Bridging Theory and Practice Rouba Ibrahim, University College London, MS&I department, UCL, Gower Street, London, WC1E 6BT, United Kingdom, Achal Bassamboo We propose a new, practice-driven, correlation-based framework to assess the relative accuracy of static and dynamic delay announcements. For a dynamic announcement, we consider the delay of the last customer to have entered service. Our approach combines queueing-theoretic analysis and an empirical study of real-life data.

219