Informs Annual Meeting 2017

MB70

INFORMS Houston – 2017

MB70

NE, Atlanta, GA, 30319, United States, yanhao@gatech.edu, Chen Zhang, Jianjun Shi This paper presents a dynamic subspace learning method for multivariate functional data modeling. We assume different functions come from different disjoint subspaces, and only functions of the same subspace have cross- correlations with each other. These subspaces can be learnt as a sparse regression problem. By allowing but regularizing the regression change over time, we can further describe cross-correlation dynamics. The model can be efficiently estimated by the fast iterative shrinkage-thresholding algorithm, and the features of every subspace can be extracted using the smooth multi-channel functional PCA. Numerical studies demonstrate the efficiency of the proposed methodology. 3 - Minimizing Negative Transfer of Knowledge Through Scalable and Regularized Cokriging Models Raed Al Kontar, University of Wisconsin-Madison, 1513 University Avenue, Room 3255, Madison, WI, 53706, United States, alkontar@wisc.edu, Shiyu Zhou Multivariate Metamodeling and the estimation of functional data is specifically challenging due to the large number of parameters to be estimated, high computational complexity and the negative transfer of knowledge which may occur if different functional output have no commonalities. In this paper, we present a scalable approach to simultaneous estimation of functional data based on the multivariate Gaussian process. Specially, we introduce a regularized process which allows sequential estimation of parameters and minimizes the negative transfer of knowledge between uncorrelated outputs. 4 - Point Cloud Data Analysis for Process Modeling and Optimization Hao Yan, 3525 Highgrove Way NE, Atlanta, GA, 30319, United States, yanhao@gatech.edu, Kamran Paynabar, Massimo Pacella Advanced 3D metrology technologies such as Coordinate Measuring Machine (CMM) and laser 3D scanners have facilitated the collection of massive point cloud data, beneficial for process monitoring, control and optimization. However, due to their high dimensionality and structure complexity, modeling and analysis of point clouds is a challenge. In this paper, we utilize techniques developed in multilinear algebra and propose a set of tensor regression approaches to model the variational patterns of point clouds and link them to process variables. The performance of the proposed methods is evaluated through simulation and a real case study of turning process optimization. 371D 11:00 - 11:45 Simio/11:45 - 12:30 AMPL Invited: Vendor Tutorial Invited Session 1 - 11-11:45 Simio Renee Thiesing, Simio LLC, Louisville, KY, United States, rthiesing@simio.com Simio is a premier simulation and scheduling software that allows you to expand traditional benefits of simulation to improve daily operations. In this tutorial, we will demonstrate Simio’s 3D rapid modeling capability to effectively solve real problems. Explore how a single tool can be used to not only optimize your system design, but also provide effective planning and scheduling. Come explore the Simio difference and see why so many professional and novice simulationists are changing to Simio. 2 - Developing Optimization Applications Quickly and Effectively through Algebraic Modeling in AMPL Robert Fourer, AMPL.Optimization Inc., 2521 Asbury Ave, Evanston, IL, 60201, United States, 4er@ampl.com Can you negotiate the complexities of the optimization modeling lifecycle, to deliver a working application before the funding runs out or the problem owner loses interest? Algebraic languages streamline the key steps of model formulation, testing, and revision, while offering powerful facilities for incorporating models into larger systems and deploying them to users. This presentation introduces algebraic modeling for optimization by carrying a single illustrative example through multiple contexts, from interactively evolved formulations to scripted iterative schemes to embedded applications. We feature the new Python API for AMPL, and the QuanDec system for creating web-based collaborative applications from AMPL models. MB69

371E Data Mining Contributed Session Chair: Safak Yakti, Binghamton University, Binghamton, NY, United States, syakti1@binghamton.edu 1 - A Classification Based Method to Detect Epilepsy using EEG Signals Eda Karacaoglan, Erciyes University, Kayseri, Turkey, edakaracaoglan@erciyes.edu.tr, Mahmut Tokmakci, Fatma Uysal It is important to diagnosis epilepsy correctly. Because EEG signals are dynamic and nonstationary, it is very hard to diagnosis epilepsy exactly. In literature there are some studies to detect epilepsy with the help of data mining methods. In this study, it has been tried to detect epilepsy by classifying Electroencephalogram (EEG) signals. For this purpose a new method is developed and applied on test data. The results are presented. 2 - A Classification Based Method to Detect Stress using ECG Signals Fatma Uysal, Research Assistant, Erciyes University, Engineering Faculty, Biomedical Engineering Department, Kayseri, 38039, Turkey, fatmauysal@erciyes.edu.tr, Mahmut Tokmakçı, Eda Karacao lan Stress is one of the serious factors for inducing many physical illnesses such as heart attacks, cognitive dysfunctions and depression. Electrocardiogram signal (ECG) which reflects the electrical activity of the heart ensures useful information in detection human stress. This work consists of five steps including stress- inducing test, data collection, signal preprocessing, feature extraction and classification. The aim of this study is to classify the stress using feature extraction algorithms and machine learning algorithms. 3 - Could Wellness Programs Help Decrease the Health Costs? M. Gabriela Sava, Clemson University, College of Business, To reduce their total healthcare costs, private companies are starting to invest more in preventive care. Wellness programs are one strategy to try to minimize the total healthcare costs. We use data from one wellness program to analyze how employees’ life style characteristics could predict future healthcare costs faced by the employer. 4 - Can Analytics Help Improve the Implementation of Preventive Care Programs? M. Gabriela Sava, Clemson University, College of Business, In order to improve public health and to reduce healthcare costs, the U.S. Preventive Task Force has developed preventive programs for chronic diseases, such as colorectal cancer. We suggest how analytical methods might be used to help increase the level of implementation of preventive screening for colorectal cancer, by improving healthcare providers’ understanding of patients’ preferences with respect to the set of screening options available. Improved understanding enables healthcare providers to propose “personalized” medical paths that should increase the likelihood of patients’ compliance. 5 - Data-driven Patient Scheduling on Asthma Intervention Considering Air Pollution Luo Li, Professor, Sichuan University, Business School, Chengdu, 610064, China, luolicc@163.com, Fengyi Zhang, Wei Zhang The aim of this study is to provide a framework of data-driven asthma patient scheduling on intervention with minimized readmissions . We first identified corresponding factors (especially on air pollution) that affect readmission and discharge process by accessing health insurance database; then we figured out the Markov transition process of readmission and discharge under each circumstance; finally, we constructed a MDP (Markov decision process) model, in which the penalties of Markov states are non-monotonic.Due to the fact that the optimality only exits under certain conditions, we proposed heuristics to achieve near- optimal and analyzed their performance. 145 Sirrine Hall, Clemson, SC, 29634, United States, msava@clemson.edu, Lawrence Fredendall, Kristin Scott 145 Sirrine Hall, Clemson, SC, 29634, United States, msava@clemson.edu, Luis G. Vargas, Jerrold H. May, James G. Dolan

189