Statistical Computing Research Environment

University Of Iowa, Iowa City IA

Investigators

Abstract

The Department of Statistics and Actuarial Science at the University of Iowa will purchase computer equipment (an Atipa 64-bit Dual-Core Opteron Beowulf cluster for intensive computation), which will be used for the following research projects: Extensions and applications of the smoothly mixing regression model (John Geweke); Regularization methods for high-dimensional regression models, with applications to survival analysis, microarray normalization and ROC classification (Jian Huang); Parallel statistical computations in R, and Bayesian analysis of PET imaging data (Luke Tierney); Bayesian models and parallel and grid-based computing strategies for high-dimensional spatial and spatiotemporal data (Jun Yan and Mary Kathryn Cowles). These projects will make substantive contributions in Bayesian econometrics; statistical genetics; statistical research, practice, and education; brain imaging; and environmental statistics. For education, the new cluster will be used for graduate courses on computational statistics and interdisciplinary seminar courses on environmental informatics, as well as in thesis research by individual graduate students. The cluster will be integrated into the University of Iowa's research computational Grid (HawkGrid), which is a node of the Open Science Grid. Thus, the cluster will contribute to broader research on Grid computing methodology, and, when not being used to capacity by researchers in the Statistics and Actuarial Science Department, it will be a research computing resource for the greater University of Iowa community and for users of the Open Science Grid. The last decade has witnessed an explosion of available data -- from satellite images and medical images to remote sensing data to databases of customer information. This phenomenon has been accompanied by an increase in the complexity of questions that people in government, business, and the sciences need answered. Five researchers from the Department of Statistics and Actuarial Science at the University of Iowa are developing and applying statistical computing methods to use massive, complex data to answer real-world questions. John Geweke's work addresses socio-economic issues, such as what factors drive the price of gasoline. Jian Huang's work involves collaboration with medical researchers in the Cancer Center at the University of Iowa and will contribute to the development of more effective methods for using genetic information in the diagnosis and treatment of cancer. Luke Tierney develops methods to help people understand data by exploring it visually as well as statistically. This work will assist educators, researchers, and users of data in business, government, and any other fields. Tierney also is involved in the statistical side of brain imaging using PET technology. Jun Yan and Mary Kathryn Cowles are working to improve statistical computing strategies for data measured over space and time, and are using their methods to study changes in the available water supply in the western United States and levels of radon gas (a risk factor for lung cancer) in buildings.