2016 INFORMS Annual Meeting Program

TA70

INFORMS Nashville – 2016

TA68 Mockingbird 4- Omni Graph Analytics for Complex Systems Sponsored: Quality, Statistics and Reliability Sponsored Session Chair: Hoang Tran, Texas A&M University, College Station, TX, United States, tran@tamu.edu Co-Chair: Satish Bukkapatnam, Texas A&M University, College Station, TX, United States, satish@tamu.edu 1 - Predicting Community Structure In Dynamic Networks: A Case Of Online Educational Networks Yi-Shan Sung, Penn State University, yqs5097@psu.edu, Soundar Kumara Community structure points to structural patterns in a network and reflects functional associations between entities. However, it is challenging to obtain timely updates of communities in a dynamic network in which changes are frequently introduced over time. We develop a model to predict community structure by integrating link prediction with community detection algorithms. We test the model efficacy using the data from nanoHUB.org, which is an online educational platform for science and engineering in nanotechnology. Predicting community structure in nanoHUB networks will help in developing an efficient recommendation system for the nanoHUB users and optimizing the resource allocation. 2 - Detecting Changes In Complex Systems Via Network Inference Hoang M Tran, Texas A&M University, College Station, TX, United States, tran@tamu.eud, Satish Bukkapatnam We propose a network based method to do change detection in transient complex systems. This is based on our approach to infer spurious-link-free network structures from time series. A spectral graph based method is used to detect process changes from these networks. 3 - Graph Reconstruction From High-dimensional Systems Of Additive Differential Equations Ali Shojaie, University of Washington, Seattle, WA, United States, ashojaie@uw.edu, Shizhe Chen, Daniela Witten We consider the task of learning a dynamical system from high-dimensional time- course data. We model the dynamical system non-parametrically as a system of additive ordinary differential equations. Most existing methods for parameter estimation in ordinary differential equations estimate the derivatives from noisy observations. This is known to be challenging and inefficient. We propose a novel approach that does not involve derivative estimation. We show that the proposed method can consistently recover the true network structure even in high dimensions, and we demonstrate empirical improvement over competing approaches. 4 - Modeling And Change Detection Of Dynamic Network Data By A Network State Space Model Na Zou, Texas A&M University, College Station, TX, 77845, United States, nzou1@tamu.edu, Jing Li Dynamic network data widely exist in social, biological, and engineering domains. There are two types of variability in dynamic network data: variability of natural evolution and variability due to assignable causes. Accurate and timely change detection from dynamic network data is important. Change detection is a classic research area in Statistical Process Control (SPC) and various approaches have been developed for dynamic data in the form of univariate or multivariate time series, but not in the form of networks. We propose a Network State Space Model (NSSM) to characterize the natural evolution of dynamic networks and integrate the NSSM with SPC for change detection.

understand Japanese real estate prices comprehensively. Our analysis reveals that the financial asset prices and conventional hedonic variables serve as the major determinants of Japanese real estate prices. 2 - Ensemble Model For U. S. Stock Major Index Prediction Using Economic Factors With Interactive Visualization Yao-Te Tsai, Post-Doctoral Fellow, Auburn University, Auburn, AL, 36849, United States, yzt0007@auburn.edu, Bin Weng, Fadel Megahed, James Barth The accuracy of the stock market prediction has been an attractive topic for researchers and public. However, it has still been remaining one of the most challenging tasks due to the non-linearity and non-stationary of the time series data. Our objective is to discover information and trends from macroeconomic perspectives to provide a foundation for the future stock market predictive model development. we investigate how macroeconomic factors that drive the U.S. major stock market index by applying the ensemble model. We also determine if the index of each sector would be driven from different factors. The last task is to predict the stock market index based on our variable selection. 3 - Capital Growth With Recycling And The Environmental Kuznets Curve Fouad El Ouardighi, Professor, ESSEC Business School, Avenue B Hirsch BP 105, Cergy Pontoise, 95021, France, elouardighi@essec.fr We investigate how the relationship between capital growth and pollution accumulation is affected by the source of pollution, that is, either production or consumption. We are interested in polluting waste that cannot be naturally absorbed, but for which recycling efforts are made to avoid massive accumulation with harmful consequences in the long run. We distinguish the cases where recycling efforts are capital-improving or capital-neutral. Based on both environmental and social welfare perspectives, we determine how the influence of the pollution source on capital growth and polluting waste accumulation is affected by the fact that recycling is capital-improving or capital-neutral.

TA70 Acoustic- Omni Transportation, Rail II Contributed Session

Chair: Yalda Khashe, University of Southern California, 3230 Overland Ave. APT 312, Los Angeles, CA, 90034, United States, khashe@usc.edu 1 - Train Timetable Based Integer Programming Model For Passenger Assignment Problem In Congest Urban Rail Line

Si Ma, Associate Professor, Southwest Jiaotong University, Chengdu, China, masi@home.swjtu.edu.cn, Gongyuan Lu, Lin Wang

We optimized passenger assignment problem (PAP) considering passenger waiting time, platform and car capacity to maximize transportation capacity in congest urban rail line. Using passenger_flow-Train_path network to integrate passenger behavior and timetable based train movement in space and time dimension, the PAP is modeled as a maximum flow problem with multi-sources and multi-sinks. In the real-world case of Chengdu urban rail line 1, the integer programming model is solved efficiently by commercial solver. 2 - Face Recognition Based Ticket Checking Solution In Speeding Train Kui Yang, PH.D. Candidate, Southwest Jiaotong University, Chendu, 610031, China, ykylw@my.swjtu.edu.cn, Gongyuan Lu, Haifeng Yan It is a big challenge to check rail ticket in almost every passenger section, due to great workload and passenger inconvenience. Integrated with ticket section, a face recognition based ticket checking solution is presented for strict and efficient checking. This visual-aided solution can automatically identify checking candidate in different section, avoid missing or multiple checking in the whole journey. 3 - Critical Systems Management Issues Of Implementing The Positive Train Control Technology In A Regional Railroad Yalda Khashe, University of Southern California, 3230 Overland Ave. Apt 312, Los Angeles, CA, 90034, United States, khashe@usc.edu Positive Train Control (PTC) is a generic term referring to a range of fully integrated technologies that overlay existing safety systems to prevent train-to- train collision and improve worker safety. One of the challenges that railroad industry is facing for implementing PTC is the complications of introducing this new technology to an already existing system and its effect on the technological, organizational and human subsystems and their interactions.

TA69 Old Hickory- Omni Economics IV Contributed Session

Chair: Fouad El Ouardighi, Professor, ESSEC Business School, Avenue B Hirsch BP 105, Cergy Pontoise, 95021, France, elouardighi@essec.fr 1 - A Note On Real Estate Pricing With Exogenous Variables Hiroshi Ishijima, Professor, Chuo University, 1-18 Ichigaya-tamachi Shinjuku, Tokyo, 1628478, Japan, hiroshi.ishijima.jp@gmail.com, Akira Maeda We develop a pricing model of real estate that incorporates conventional hedonic attribute variables of real estate as well as exogenous variables, namely financial asset prices; this model is based on a theoretical pricing model that we, fundamentally develop. Specifically, our model features a pricing kernel expressed as the product of a cash-flow pricing kernel (stochastic discount factor) and a hedonic pricing kernel. Furthermore, we conduct an empirical analysis to

257