This is an interdisciplinary area utilizing concepts from Statistics, Computer Science, Graphics, and Numerical Mathematics.
Some examples of our work are the following:
|Methodology for analysis of huge datasets by extracting and classifying features worthy of further investigation: paper.pdf (270 Kbytes), paper1 (2,642 Kbytes), paper2 (2,771 Kbytes), paper3 (184 Kbytes), prototype software. Several methods are developed. Among them, local modeling with global analysis can be viewed as a generalization of chaotic time series analysis. The "fireworks" on the right are the result of cluster analysis applied to parameters of local models on an atmoshperic radiation measurement time series. Such analysis can identify and classify unusual segments of large data series. In most cases, each cluster has an application-specific interpretation because the local models are application-specific.|
|Methods for estimating cumulative radiation doses using historical data that are subject to bias and uncertainty (Radiation Research (1997)). We build likelihood functions that represent the forward measurement process. Bayes theorem and some computation then solves the inverse problem. Finally, discrete Fourier transorms are used to combine the solutions into a cumulative dose distribution.|
|An algorithm for data-driven global model search with respect to a model optimality criterion within a large user defined class of models (viewgraph, Computational Statistics and Data Analysis (1993), software). A model optimality criterion provides a linear ordering for models in a large hierarchical class of models. An algorithm is developed for generating the class and navigating within the class of models. Then, a small subset of best models with respect to the optimality criterion can be efficiently found using a bounding process. More conventional analysis of the small and manageable subset can then select a single model that is consistent with application knowledge.|
The Computing facilities available for our work in computational statistics include those of the Computational Mathematics and Statistics Section (CMSS), which is within the Computer Science and Mathematics Division (CSM) which houses many networked high performance workstations as well as parallel computers. Within CSM, there is an Advanced Visualization Research Center that has a number of high performance visualization workstations and other related high resolution graphics equipment. Network access is also available to supercomputers of the Center for Computational Sciences, which is housed in a nearby building. A high speed link connects CMSS with the University of Tennessee (UT) Computer Science Department and the Joint Institute of Computational Science to facilitate the use of computers at both sites.