Interdisciplinary Workshop for Undergraduate Students May 22-27, 2016 Plenary Talk Richard Smith Speaker Titles/Abstracts Statistical Reasoning in Public Even small datasets can pose complex problems of statistical interpretation. This talk arises from a query that was made to me by a journalist, whose question initially seemed very simple. Closer study showed, however, that it could be interpreted in different ways that led to different conclusions depending on how the data was analyzed. This talk will present the problem and data, and will discuss different ways of thinking about it and what statistical conclusions they lead to. The broader message is that data science is not just about computation: thinking carefully about the objectives of a statistical analysis, and developing procedures appropriate to those objectives, are a critical part of the process. I. Project-I [NBA] Lucas Mentch and Benjamin Risk Project Abstracts Creating a Data-Based 2016 NBA Mock Draft At the time of this workshop, the 2016 NBA draft will be just one month away and plenty of analysts and websites will have created mock drafts predicting where players will be drafted. For this project, we'll create historical datasets including information on college performance, individual measurements, and previous draft results in an effort to predict NBA performance of current college basketball players. We'll then transform this into a projected draft order and compare to other popular mock drafts. II. Project-II [Dolphin] Adam Jaeger, Claire Johnson, UNC Behavior of Bottlenose Dolphins (Tursiops Truncatus) in Roanoke Sound, North Carolina
Bottlenose dolphins are essentially the top predator in Roanoke Sound, making their presence in the area an indicator of ecosystem health. If we know when and why they migrate, this will help researchers better understand what an unusual year might look like and give clues as to what may be causing a decrease or increase in seasonal population. The goal is to examine how various environmental factors predict behaviors of bottlenose dophins in the Northern North Carolina Estuarine System (NNCES) stock in Roanoke Sound, North Carolina. A primary question of interest is how water temperature relates to the presence of dolphins, the number of calves present and the behavior of the dolphins. Furthermore there may be additional confounding variables not directly related to water temperature that may also affect dolphin location and behavior. Our dataset was composed of dedicated surveys and opportunistic sightings collected by the Outer Banks Center for Dolphin Research (OBXCDR) from June 2008 to October 2015 (n=1,513). III. Project-III [DTI] Benjamin Risk and Zhengwu Zhang Diffusion Tensor Imaging Analysis Diffusion tensor imaging (DTI) is a technique for measuring the diffusion of water in brain tissue. Water has different diffusion properties in different types of tissue, where diffusion is more restricted in white matter than gray matter regions. DTI can be used to detect white matter tracts, which act as highways for neural activity by connecting regions of the brain that may be spatially distant. The goal of this project is to determine whether diffusion properties are related to cognitive ability. The data comprise diffusion measurements along dozens of locations in the corpus callosum, which is a large white matter structure that connects the hemispheres of the brain. IV. Project-IV [Fingerprint] Hoang Duy Thai, Zhengwu Zhang and Lucas Mentch A Correlation Based Approach to Quality and Noise in Crime Scene Fingerprint It is known that different sensors produce different images.thus, estimation of image quality is a challenging problem in general, especially for rolled and plain fingerprints constructed by well-defined patterns. There are usually two kinds of errors in fingerprint images, namely error produced by sensors and error produced by users, e.g. smudges in a wet fingerprint. The error produced by sensor is mathematically qualified as highly correlated random variable in small scale and error produced by users as smudges is considered as large scale noise. Based on these information and together with some fundamental tools in image analysis (e.g. image segmentation, harmonic analysis and minimization, etc), the purpose of this project is to analyze statistical properties of these noise for estimation of image quality.
Photos and Bios of Speakers Adam Jaeger Adam Jaeger is currently working in the Challenges in Computational Neuroscience at. His research interests are in topological data analysis, nonparametric likelihood theory and climatology. Claire Johnson Claire graduated from UNC Chapel Hill in May 2016 with majors in Environmental Studies (Ecology) and Geography. While living in the Outer Banks in the fall of 2015 to study North Carolina s small oyster aquaculture industry, she interned with the Outer Banks Center for Dolphin Research (OBXCDR). At OBXCDR, she studied the relationship between dolphin abundance and water temperature in Roanoke Sound and is working to publish the results. Claire is currently a Research Technician at the UNC Coastal Studies Institute working on the geospatial analysis for a project investigating the potential for energy generation using the natural salinity gradients of the water in the Albemarle-Pamlico Estuarine System. Lucas Mentch Lucas Mentch is an Assistant Professor in the Department of Statistics at the University of Pittsburgh. During the 2015-2016 academic year, Lucas was on leave participating in the program on Forensic Science.
Michael Rappa Michael Rappa is the founding director of the Institute for Advanced Analytics and a member of the faculty in the Department of Computer Science at North Carolina State University. As head of the Institute, he leads the nation s first Master of Science in Analytics as its originator and principal architect. Before joining NC State as Distinguished University Professor in 1998, for nine years he was a professor at the Massachusetts Institute of Technology. Appointed the inaugural Goodnight Director in 2015, his current endowed position is named in honor of NC State s distinguished alumnus and prominent business leader Dr. James Goodnight. Benjamin Risk Benjamin Risk is a postdoc in the program in Challenges in Computational Neuroscience. He completed his PhD in Statistics from Cornell University in the summer of 2016. His research interests include statistical methodology for the analysis of neuroimaging data, independent component analysis, spatiotemporal processes, and ecology. Richard Smith Richard L. Smith is Mark L. Reed III Distinguished Professor of Statistics and Professor of Biostatistics in the University of North Carolina, Chapel Hill. He is also Director of the Statistical and Applied Mathematical Sciences Institute, a Mathematical Sciences Institute supported by the National Science Foundation. He obtained his PhD from Cornell University and previously held academic positions at Imperial College (London), the University of Surrey (Guildford, England) and Cambridge University. His main research interest is environmental statistics and associated areas of methodological research such as spatial statistics, time series analysis and extreme value theory. He is particularly interested in statistical aspects of climate change research, and in air pollution including its health effects. He is a Fellow of the American Statistical Association and the Institute of Mathematical Statistics, an Elected Member of the International Statistical Institute, and has won the Guy Medal in Silver of the Royal Statistical Society, and the Distinguished Achievement Medal of the Section on Statistics
and the Environment, American Statistical Association. In 2004 he was the J. Stuart Hunter Lecturer of The International Environmetrics Society (TIES). He is also a Chartered Statistician of the Royal Statistical Society. Hoang Duy Thai I am currently a Postdoctoral Fellow at Statistical and Applied Mathematical Science Institute () in the Forensics program with Prof. D. Banks (Duke), L. Stefanski (NCSU) and R. Smith (UNC). I did my PhD degree in Institute for Mathematical Stochastics in University of Goettingen, Germany, under supervision of Prof. Axel Munk. There, I was a member of the DFG Graduate Program 1023 - Identification in Mathematical Models. My master degree was in Bio-Mechatronics Engineering, Sungkyunkwan University, South Korea where I worked as a research assistant (mainly for image analysis) in Robotics and Vision lab and my position was funded by the government of South Korea. My bachelor degree was in Control Theory in the honor program of Electrical-Electronics Engineering from Ho Chi Minh City University of Technology, Vietnam. My interests are mathematical imaging in inverse problem and control theory, including harmonic analysis, PDE, optimization and statistics. Thomas P. Witelski My primary area of expertise is the solution of nonlinear ordinary and partial differential equations via perturbation methods. Using asymptotics along with a mixture of other applied mathematical techniques in analysis and scientific computing I study a broad range of applications in physical systems. Focuses of my work include problems in viscous fluid flow, industrial applications, flow in porous media, mathematical biology. Through my research I am working to extend the understanding of nonlinear diffusion processes in physical systems. Studying problems in a range of different fields has given me a unique opportunity to interact with a diverse set of collaborators and to transfer analytic techniques across the traditional boundaries that separate fields. Zhengwu Zhang I am currently a Postdoctoral Fellow at Statistical and Applied Mathematical Science Institute () affiliated with the Challenges of Computational Neuroscience (CCNS) program. My advisors are Prof. Hongtu Zhu (UNC Chapel Hill) and Prof. David Dunson
(Duke). I am working with them on various problems in the computational neuroscience: structural connectivity, brain network, image and shape analysis. I earned my PhD in Statistics from Florida State University (FSU), under the supervision of Prof. Anuj Srivastava. I was working in developing statistical methods for functional and shape data, e.g. 2D contours, surfaces, signals, densities, and trajectories on non-linear manifolds. I received the B.S. degree in Engineering from South China University of Technology (SCUT) in 2008, where I was a participant of Talented Student Program.