Informs Annual Meeting 2017

MB39

INFORMS Houston – 2017

MB39

MB41

352D Innovative Technologies Cluster: Analytics Sponsored Session 1 - Crowd-Sourcing Analytics with Datasplash Lorens Helmchen, George Washington University, Washington, DC, United States, helmchen@gwu.edu We introduce Datasplash, a new web-based analytics application that allows any- one to load, explore, and share their data quickly, easily, and securely on any device anywhere — without downloading and installing software or learning a programming language. We show how even users with no analytics training can spot salient and significant associations, assess model fit, and compute predic- tions. We also show how Datasplash can be used to recognize and remedy more advanced model-fitting challenges, such as non-linearities, outliers, confounding, and collinearity. We illustrate these features by applying them to case studies from the hospitality and health care industries. 2 - Communicating with the New Machine: Human Insight at Machine Scale Kris Hammond, Narrative Science, Chicago, IL, United States, khammond@narrativescience.com In this talk, I will outline what we are with regard to data analytics and how the technology of automatic narrative generation from data plays the crucial role of bridging the gap between the Big Data world of numbers and symbols and our need for understandable insights. I will dive into use cases from business, educa- tion and everyday life to show how the power of automatically generated narra- tives can provide us all with the insights that are still trapped in the wealth of data we now control. 352E Daniel H. Wagner Prize Competition I Invited: Daniel H. Wagner Prize Competition Invited Session Chair: Patricia Neri, SAS Institute, Inc., 104 Grandtree Ct., Cary, NC, 27519, United States, patricia.neri@sas.com 1 - The Inmate Assignment and Scheduling Problem and its Application in the PA Department of Corrections Tamás Terlaky, Lehigh University, 200 W. Packer Ave, Bethlehem, PA, 18015, United States, terlaky@lehigh.edu, Mohammad Shahabsafa, Chaitanya Gudapati, Anshul Sharma, Louis J. Plebani, George R. Wilson, Kristofer B. Bucklen The inmate assignment project, in close collaboration with the PA Dept. of Corrections, took five years from start to successful implementation. Our novel Inmate Assignment Decision Support System (IADSS) is designed with the main goal of simultaneously, and system-wide optimally, assigning the inmates to the correctional institutions. IADSS includes a new hierarchical multi-objective MILO model, which accurately describes the inmate assignment problem. This is the first time that OR methodologies have been used to optimize the operations, and built into the routine business practice, of a correctional system, thus it opens a rich and untouched area for the application of OR. 2 - Forecasting and Optimization Models for Audience Targeting on Television Wes Chaar, Turner Broadcasting System, Inc., 106 N.Denton Tap Road, # 210-273, Coppell, TX, 75019-2148, United States, wscemail2002@yahoo.com, J. Antonio Carbajal, Peter Alexander Williams At Turner, we have developed and implemented innovative, integrated forecasting and optimization models that forecast audiences in the 24/7 programming schedule, generate media plans across all Turner’s networks, and schedule commercials holistically. These methods power Turner’s audience targeting solutions: TargetingNOW and AudienceNOW, which have produced significant sales and advertisement efficiencies for Turner and its clients, and moved the industry forward in the emerging audience targeting landscape. MB40

352F Health Care, Modeling and Optimization Contributed Session Chair: Danielle Ripsman, Waterloo Ontario, ON, Canada, daripsman@uwaterloo.ca 1 - Joint Optimization of Appointment and Real-time Schedule in the Presence of Walk-ins, Waiting Time Targets and No-shows Xingwei Pan, Shanghai Jiao Tong University, No. 800 Dongchuan Road,Minhang District, Shanghai, 200240, China, 1103951776@sjtu.edu.cn Clinics give the highest priority to appointed patients, incurring long waiting of walk-ins. To balance the waiting between the appointed and walk-ins, this paper jointly optimizes the appointment and real-time schedule by considering appointed patients’ waiting time target (WTT) and no-shows. A finite horizon Markov Decision Process model is proposed to establish structural properties of the optimal control policy. Appointment schedule is gained by combining a stochastic programming model and local search. Numerical tests show the impact of WTT and efficiency of the approach. Key words: appointment; real-time schedule; waiting time target; Markov Decision Process; Stochastic Programming 2 - Policy Design for Online Scheduling with Heterogeneous Patients and Doctors Shan Wang, Shanghai Jiao Tong University, 1954 Huashan Road, Shanghai, 200030, China, wangshan_731@sjtu.edu.cn, Huiqiao Su, Guohua Wan How to design the patients scheduling policy for the heterogeneous patients and doctors is a problem in health care management. We study the process of outpatient clinic service in China and find there exit problems in outpatient scheduling. In this paper, we show how to model this problem as an online scheduling problem, and demonstrate how to solve this scheduling problem by applying the Markov Decision Process. We design the numerical experiments to compare the performances of our policy and two policies employed in practice by simulating various scenarios; the numerical results show that our policy has the best performance under all scenarios, especially when the system has a heavy workload. 3 - Nurse Scheduling with Shift Preferences using Goal Programming Esra Agca Aktunc, Kadir Has University, Kadir Has Cd. Cibali, Istanbul, 34083, Turkey, esra.agca@khas.edu.tr, Elif Tekin Fair workload and job satisfaction should be taken into account in nurse scheduling to improve the quality of healthcare. In this study, a monthly nurse scheduling model is developed based on labor regulations, hospital management requirements, and nurse preferences. A multi-objective integer program is formulated and solved as a goal programming model to avoid disproportional shift assignment, improve job satisfaction of nurses, and reduce medical errors caused by fatigue. Resulting schedules for various shift preference scenarios are compared to show the improvement in the schedule quality in terms of performance measures such as distribution of preference violations among nurses. 4 - A Polynomial-time Approximation Scheme for Sequential Batch-testing of Series Systems Yaron Shaposhnik, Assistant professor, University of Rochester, Simon Business School, CS.H.3-343, 500 Joseph C. Wilson Blvd., Rochester, NY, 14627, United States, yaron.shaposhnik@simon.rochester.edu, Danny Segev We study a recently-introduced generalization of the classic sequential testing problem for series systems, consisting of multiple stochastic components. The conventional assumption in such settings is that the overall system state can be expressed as a boolean function, defined with respect to the states of individual components. However, unlike the classic setting, rather than testing components separately, we allow aggregating multiple tests to be conducted simultaneously, while incurring an additional set-up cost. We devise a PTAS for the sequential batch testing problem, thereby significantly improving on the constant-factor approximation of 6.829+eps due to Daldal et al. 5 - Personalized Fatigue-aware Scheduling of Physical Activity using Stochastic Programming Alla Kammerdiner, New Mexico State University, 7474 Sierra Luz Dr, Las Cruces, NM, 88012, United States, alla@nmsu.edu Sedentary people benefit from physical activity. Yet, overexertion from an activity induces fatigue, which can discourage the following physical activity and long- term adherence to exercise. In this study, we develop and formulate new stochastic programming problems for scheduling bouts of physical activity in sedentary people to maximize its health benefit while limiting overexertion. The models are formulated to maximize the time or amount of activity while constraining the chance of exercising past fatigue onset, which could have a negative impact on adhering to exercise. We present the solution algorithms and illustrate the application of the developed optimization problems on real data.

178