Informs Annual Meeting 2017

SA70

INFORMS Houston – 2017

SA69

SA70

371D Elements Toward a Smart City Sponsored: Energy, Natural Res & the Environment Environment & Sustainability Sponsored Session Chair: Victoria Chen, University of Texas, vchen@uta.edu 1 - Design and Analysis of Computer Experiments Approach for Green Building Shirish Rao, The University of Texas-Arlington, Arlington, TX, United States, shirish.rao@mavs.uta.edu Building structures have significant impact on environment and energy consumption. Buildings can be designed so that their energy consumption is reduced by using energy efficient materials. The “cradle-to-grave” cycle is a criterion for the environmental effects of materials, and minimizing this effect along with energy consumption is of great interest. The goal of the research is to compare different experimental designs and statistical modeling methods to help inform our approach for a multivariate, multi-response green building framework. The results of the experiments are studied by using Treed regression and Multi Adaptive Regression Splines techniques to find influential factors. 2 - Using Approximate Dynamic Programming to Control an Electric Vehicle Charging Station System Victoria C.P. Chen, University of Texas-Arlington, Industrial, Manufacturing, & Systems Engr., Campus Box 19017, Arlington, TX, 76019-0017, United States, vchen@uta.edu, Ying Chen, Jay Michael Rosenberger, Wei-jen Lee Electric vehicle charging stations system assumed to locate in Dallas Fort-Worth area integrate wind power, solar power and main electricity grid to supply electricity for the demand from the electric vehicles. Design and analysis of computer experiments based infinite-horizon approximate dynamic programming (ADP) algorithm is used to generate a high-quality policy to control this 31 dimensional system over a continuous-state space. A specified 45-degree line correspondence stopping rule is utilized to identify the high-quality ADP policy at early DP iterations. The results demonstrate with the selected ADP policy, the system has a better control performance than using the greedy policy. 3 - Optimizing Location of Electric Vehicle Charging Stations Ukesh Chawal, The University of Texas at Arlington, Arlington, TX, United States, ukesh.chawal@mavs.uta.edu, Jay Rosenberger, Victoria Chen, Wei-Jen Chen, Raghavendra Punugu, Mewan Wijemanne Due to the economic and environmental concerns associated with fossil fuels and the growing need for sustainability, the use of electric vehicles (EVs) is a viable solution. In this paper, a Mixed-Integer Linear Programming (MILP) model and a Design and Analysis of Computer Experiments (DACE) based system design optimization approach have been used to optimize the locations of the EV Charging stations and the number of slots to be opened to maximize the profit, meeting the customers demand. 4 - A Simulation Framework for Police Patrol Deployment: A Decision Support Tool Khan Md. Ariful Haque, The University of Texas at Arlington, Arlington, TX, United States, khanmdariful.haque@mavs.uta.edu Evaluating dynamic deployment strategies and determining the best one is a real challenge to a police department without reforming the current practice that costs money and time without any result. A flexible discrete-event simulation framework has been proposed that can facilitate any police department in city level for evaluating possible dynamic policing strategies without spending significant resources. From available data, estimated input parameters considering randomness of Call arrivals and service times of different priority levels are using into the model that will enable police department in decision making based on the output.

371E Big Data Contributed Session Chair: Harshad Rai, University of Illinois at Urbana-Champaign, Champaign, IL, United States, hrrai2@illinois.edu 1 - Looking Inside Your Shopping Bag the use of Retail Data to Capture Health Lifestyles lichung Jen, National Taiwan University, Taipei, Taiwan, lichung@ntu.edu.tw, YiChun Liu When a patient goes to a clinic, the doctor can only see the results from all the medical examinations and then infer what is wrong about his/her health without knowing what exactly is the behavior of his/her daily nutrition intake. Due to the recent development of Big Data technology, the transactions data collected by retailers can reveal the details of consumer purchase behavior illustrating their health lifestyle. In this study, we construct an index of calories activity (CAI) of food consumption based on the customer transaction database of a supermarket. Through the index, we are able to identify customers with health risks. 2 - The Optimal Bootstrap and its Applications Bradley E. Sturt, Massachusetts Institute of Technology, Cambridge, MA, United States, bsturt@mit.edu, Dimitris Bertsimas The bootstrap is a randomized method for calculating quantities, such as confidence intervals (CI), directly from data. Since the variability from randomization can lead to inaccurate outputs, we take a deterministic approach. In this talk, we present the first efficient deterministic approximation algorithm (FPTAS) for the bootstrap method. In experiments, the algorithm quickly produces CIs that are substantially more accurate and trustworthy than those from traditional methods. Finally, we present several new computational complexity results related to the bootstrap and probability theory. 3 - Internet of Things 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. 4 - GPU Supported Large Scale Simulation Models for Influenza Pandemic Outbreaks Shalome Hanisha Anand Tatapudi, University of South Florida, 4202 E. Fowler Avenue, ENB 118, Tampa, FL, 33620-5350, United States, tatapudi@mail.usf.edu, Walter Silva, Tapas K. Das Influenza pandemics are a serious concern and researchers are trying to understand its patterns. One such tool to effectively understand the disease characteristics is through an agent-based (AB) simulation model, which is versatile, yet has computational limitations when it comes to simulating larger populations. This study integrates the flexibility of AB simulation with computational efficiency of a graphical processing unit (GPU) to create models for pandemic outbreaks in large areas comprising of hundreds of millions of people. 5 - NYC Traffic Collision Analysis to Mine Patterns in Traffic Fatalities Harshad Rai, Graduate Student, University of Illinois at Urbana- Champaign, 107, East Springfield Ave., Apt. 317, Champaign, IL, 61820, United States, hrrai2@illinois.edu, Anirudh Madhusudan This project focuses on the use of Machine Learning algorithms to analyze traffic collisions. In the future, there is likely to be a lot more data about cities. Vehicle- to-vehicle and vehicle-to-infrastructure is likely to provide transformative changes in mobility. Cities are breeding grounds for emergent behavior, so this provides a lot of interesting challenges. The DBSCAN algorithm, a density based clustering method; is used to cluster traffic fatalities to recognize possible danger zones. Spatio-temporal visualization and analysis is carried out to find patterns in traffic collisions. Random forest is used to predict the recurrence of accidents and fatalities.

27