Informs Annual Meeting 2017

MA70

INFORMS Houston – 2017

MA70

for convex optimization problems. We will specifically show how the database can be used to solve problems pertaining to one of the largest railway networks in the world. The database will store data as polytopes derived from historical and apriori structural relationships and compresses the infinite possibilities of train arrival or departure, into a very compact representation. We will show how the database can answer business use cases such as conflicts in train schedules through relational algebraic operations on the polytopes.

371E Big Data Contributed Session

Chair: Anushka Chandrababu, International Institute of Information Technology, Bangalore, Bangalore, India, anushka.babu@iiitb.org 1 - IoT Security Michael Chuang, SUNY New Paltz, 1 Hawk Drive, New Paltz, NY, 12561, United States, mikeychuang@gmail.com Internet of Things, a global infrastructure enabling services by interconnecting physical and virtual things based on interoperable information and communication technologies, has emerged in business applications in various industrial areas, and will reshape tomorrow’s business and technology development. 2 - Individual Social Behavior and Intraday Patterns in Market Movements Keli Xiao, Stony Brook University, College of Business, Stony Brook, NY, 11794, United States, keli.xiao@stonybrook.edu, Gaoshan Wang, Gaoshan Wang, Jishun Wang, Feiran Li, Xiaoyan Ding In this study, we investigate the intraday patterns of individual investors’ social behavior and the stock market dynamics with several new technical attempts including (1) developing an investor engagement index (IEI) using stock forum activity data based on the customer engagement theory, and (2) proposing a multi-source forecasting framework that combines both time series models and machine learning algorithms. With the IEI and transaction data of each stock, the relationship between IEI and stock movement is investigated and discussed. Based on the data collected from a famous Chinese stock forum, we benchmark the performance of our approach with several classic baselines. 3 - The Sequence Effects of Customer Experiences on Customer Perceived Value in Wellness-centric vs. Standard Hotel Stays: A Structural Econometric Analysis Min Kyung Lee, Clemson University, 100 Sirrine Hall, Clemson, SC, 29634, United States, minl@g.clemson.edu, Aleda Roth, Bernardo F. Quiroga, Rohit Verma We explore emerging market trend of wellness industry that provides service offerings or products reactively to healthy people to make them feel even healthier. Our work specified a structural econometric model to explore the theoretical importance of the service process sequencing for wellness-centric rooms versus exactly similar standard rooms. Specifically, we account for customers’ experiences from hotel check-in to room stay to check-out on their satisfaction and likelihood to recommend for both wellness-centric rooms and standard rooms. We demonstrate that the wellness-centric service offering produces more perceived customer value added, when benchmarked against identical rooms. 4 - Neural Network Predictions of NYC Traffic Collisions Anirudh Madhusudan, Graduate Student, University of Illinois- Urbana Champaign, 909S.Fifth Street, 120A Sherman Hall, Champaign, IL, 61820, United States, anirudh2@illinois.edu, Harshad Rai The NYPD Motor Vehicle Collision dataset offers information on the traffic collisions and fatalities that have occurred in NYC since 2012. Also since 2012, the Vision Zero initiative has been trying to incorporate road-safety measures to reduce traffic incidents. This projects uses several visualizations and machine learning algorithms including Logistic Regression and Neural Network to learn and predict the occurrences of collisions and fatalities usings the dataset. The project also studies the impact of the treatments offered by Vision Zero to alleviate the condition of New York traffic. 5 - Dependency Discovery and Quality Improvement in Multistage Manufacturing Processes Andi Wang, Georgia Institute of Technology, Atlanta, GA, United States, andi.wang@connect.ust.hk, Hao Yan, Jianjun Shi Many modern fabrication processes such as semiconductor fabrication involve a large number of stages. Massive data are generated from each stage for quality improvement. Although the correlation structure between stages can be very complex, existing methods either analyze the data acquired from each stage independently or apply a state-space model that only captures the dependency of subsequent stages. In this research, we aim at detecting and modeling the complex dependency structure of data from multi-stages processes, which will provide great opportunity for further quality improvement. 6 - Convex Model Database – An Extension to Transportation Networks Anushka Chandrababu, PhD Candidate, International Institute of Information Technology, Bangalore, 26/C, Electronic City, Bangalore, India, anushka.babu@iiitb.org, Srinivasa Prasanna We present our extensions to the Convex Model Database (CmdB) suitable for summarizing structured or unstructured big-data and used for answering queries

MA71

371F Journal of Global Optimization: INFORMS Issue Sponsored: Optimization, Global Optimization Sponsored Session Chair: Sergiy Butenko, Texas A&M University, Texas A&M University, College Station, TX, 77843-3131, United States, butenko@tamu.edu 1 - Global Optimization with Surrogate Approximation of Constraints Juliane Mueller, Lawrence Berkeley National Lab, We introduce GOSAC, a global optimization algorithm for problems with computationally expensive black-box constraints and cheap objective functions. GOSAC first finds a feasible point by solving a multi-objective optimization problem in which the constraints are minimized simultaneously. Subsequently, GOSAC aims at improving the feasible solution. We use cubic radial basis function surrogate models to approximate the constraints. We iteratively select sample points by minimizing the cheap objective function subject to the constraint function approximations. We assess GOSAC’s efficiency on cheap test problems with integer and continuous variables and two environmental applications. 2 - A Lagrangian Search Method for the P-median Problem Jiming Peng, University of Houston, Houston, TX, United States, jopeng@Central.uh.edu In this talk, we present a new method for the P-median problem. We first recast the P-median problem as a global optimization problem over the set of Lagrangian multipliers. Then we present a stochastic search method to find the Lagrangian multipliers corresponding to the optimal solution of the original P-median problem. Numerical experiments show that our method can effectively find a global optimal or very good suboptimal solution to the underlying P-median problem, especially for P-median problems with a large gap between the original problem and its Lagrangian relaxation. 3 - Strong Valid Inequalities for Boolean Logical Pattern Generation Hong Seo Ryoo, Professor, Korea University, 145 Anam-Ro, Seongbuk-Gu, Seoul, 02841, Korea, Republic of, hsryoo@korea.ac.kr, Kedong Yan 0-1 multilinear programming (MP) captures the essence of pattern generation in Logical Analysis of Data (LAD). This paper utilizes graph theoretic analysis of data to discover useful neighborhood properties among data for data reduction and multi-term linearization of the common constraint of an MP pattern generation model in a small number of stronger valid inequalities. This means that, with a systematic way to more efficiently generating Boolean logical patterns, LAD can be used for more effective analysis of data in practice. Mathematical properties and the utility of the new valid inequalities are illustrated on small examples and demonstrated with 12 real-life data mining datasets. 2453 Bonar Street, Berkeley, CA, 94702, United States, juliane.mueller2901@gmail.com, Joshua Woodbury

MA72

372A Network Optimization 5 Sponsored: Optimization, Network Optimization Sponsored Session

Chair: Shanshan Hou, University of Arizona, University of Arizona, 1127 E James E. Rogers Way, Room 323D, Tucson, AZ, 85721, United States, shanshanh@email.arizona.edu 1 - Clinic Appointment Scheduling for Integrated Practice Units Pengfei Zhang, The University of Texas at Austin McCombs School of Business, Austin, TX, 78731, United States, pz@utexas.edu, Douglas Morrice, Jonathan F.Bard Clinic appointment scheduling is very important for clinic management because it affects the providers’ time usage and patients waiting time. It is important that we have a schedule that can balance providers’ time usage and patients’ waiting. We consider a clinic appointment scheduling problem with multiple types of providers and two types of patients. The objective is to minimize the clinic closing time while restrain the patients’ waiting time to be within a certain range. We propose a new model for the problem, and develop heuristic method to solve it.

158