Time series, stochastic processes, risk analysis, artificial intelligence, econometrics
Factor models, Gaussian process, high-dimensional data, large contingency tables
My main general methods currently are in developing statistical methods within the frameworks of semiparametric and functional regression, understanding the structure of variability, and in latent variables especially as they arise in the case that important variables are measured with error and subject to excess zeros. My main application interest lies in problems of nutrition and physical activity, both at the molecular level and in the individual level. This has led me recently to considering problems of gene-environment interactions (where nutrition is the environment) and their effect on cancer, understanding the nature of dietary intake patterns in humans and the effect of those patterns on a host of diseases, and most recently, whether increasing physical activity increases the mean and decreases the variance of sleep efficiency.
Long memory time series, econometrics
My research interests fall under the general heading “applied probability and stochastic processes”. Put simply, I like to study the probabilistic structure and properties of processes that statisticians and others want to observe and analyze. For example, nonlinear time series models are being applied to many observed times series, using statistical methods that assume some type of stability – my work looked at how one can verify such stability for specific models. More recently, I and colleagues from computer science are investigating properties of dynamic random networks with an interest in optimizing some of those properties. I am also interested in questions of real analysis that are related to or devolve from statistical problems. This has included looking at the distributions of heavy-tailed random variables, which is relevant for risk theory and extreme values of data, and at the relationship between discontinuous functions and their Fourier transforms, which was of interest for function estimation.
Microarrays, bioinformatics, classification methods, statistical education
P. Fred Dahm
Measurement error models, biostatistics, econometrics
My interests include nonparametric function estimation, hypothesis testing in complex settings, time series analysis and Bayesian methods. Recently I have focused attention on inference problems involving a large number of small data sets. Suppose, for example, that the distribution of data within different small data sets is the same up to location and scale, with location and scale differing randomly from one data set to the next. I am interested in methods for estimating the distribution common to all data sets, and also the distribution of location and scale across data sets. I have considered both frequentist (minimum distance) and Bayesian methods for tackling this problem.
Another problem involving a large number of small data sets is testing the equality of distributions across all data sets. This is like the classical k-sample problem, but with the key difference that instead of fixing k and allowing sample sizes to increase without bound, I let k tend to infinity with sample sizes fixed. Doing so leads to different and interesting asymptotics in this and other inference problems.
I am also interested in simulation methods that involve generating many different models randomly rather than generating large numbers of data sets from just a few models. I call this approach BayesSim since the distribution from which models are selected is analogous to the prior distribution in Bayesian methodology.
Statistics education, consulting
Nonparametric and semiparametric methods, statistical function estimation using polynomial splines, statistical methods for longitudinal data/panel data, multivariate/functional data analysis, survival analysis, duration data, event history analysis, statistics application in business
My current methodological research interests focus on problems related to Bayesian hypothesis testing, Bayesian variable selection in ultra-high dimensional spaces, and latent variable models for ordinal and rank data analyses. In the areas of hypothesis testing and variable selection, I am particularly interested in exploring efficiencies that can be gained through the use of non-local prior densities to specify either alternative hypotheses in hypothesis testing problems or the non-null distributions of regression coefficients in variable selection problems. My research on ordinal and rank data modeling finds application in evaluating the intelligence of non-human primate species and in educational assessment.
Applied problems that currently interest me include statistical studies of the impacts of college admission policies on diversity of college campuses and graduation rates and developing more meaningful instruments to evaluate undergraduate and graduate teaching. More generally, I am interested in studying the impact that assessment procedures have on educational processes.
Finally, I am intrigued by the problem of performing inference in agent models used in sociology and psychology.
Applied statistics, statistical computing, data mining/machine learning
My research focuses on developing statistical models and methods for spatial and spatial-temporal data. In particular, I am interested in studying and modeling the nonstationary covariance structure and the nature of the spatial-temporal interactions in various environmental applications.
Regarding statistical methodology and theory, I am interested in developing parametric covariance model classes suitable for spatial and spatial-temporal processes on the surface of a sphere, geared towards data sets on a global scale (such as satellite data and numerical model outputs), and studying their characteristics. I work on various application problems, such as data assimilation (ensemble Kalman filter method), validation of climate model outputs, and analysis of surface images from geographic applications.
My research interests are mainly in Bayesian spatial statistics, with applications in the environmental sciences. As remote-sensing instruments mounted on satellites have made it possible to collect massive amounts of data on a global scale, much of my research focuses on the development of complex, flexible spatial methods that can be applied to big global datasets in a computationally feasible way. For example, I work with collaborators at NASA and NCAR on combining measurements from several satellites measuring CO2 on a global scale, on how to run related algorithms in parallel in modern distributed-computing environments, and on the real-time analysis of massive, streaming spatio-temporal datasets that are important for forecasting severe rainfall.
Spatial statistics, statistics education, consulting
My interests are in statistical challenges in astronomy data, in particular classification and prediction problems. Astronomy data sets provide a number of statistical challenges: objects are high dimensional (eg images or functions), features are often observed with heterogeneous measurement error, data sets are large, and sample selection bias is often present. I collaborate with several members of the Texas A&M Physics and Astronomy Department, including Lucas Macri and Casey Papovich. Methodologically I’m interested in classification (in particular tree based classifiers such as random forest) and measurement error.
My research is collaborative research with faculty members in entomology, animal science, and numerous other departments throughout the Texas A&M campus. My contributions consist of helping researchers design experiments, determine sample sizes, decide on appropriate models, select statistical methodology to analyze their data, and produce insightful graphs. This, hopefully, results in efficient designs, more powerful tests, and improved explanations of the results.
Bayesian hierarchical modeling, nonparametric regression and classification, bioinformatics, spatio-temporal modeling, machine learning, functional data analysis, Bayesian nonparametrics, petroleum reservoir characterization, uncertainty analysis of computer model outputs
My main research area is inference for regression and stochastic process models. I am particularly interested in flexible semiparametric methods that require much weaker model assumptions than classical methods, which are typically likelihood based. An important focus is on characterizing and constructing estimators that are efficient in the sense of the Hájek and Le Cam theory for locally asymptotically normal families. Getting to the bottom of a problem sometimes leads to a new approach that is surprisingly simple but nonetheless optimal. Several of my recent papers and projects are on efficient estimation in nonparametric and semiparametric missing data models.
H. Joseph Newton
Time series analysis, computational statistics
My research is focused on modeling dependence (covariances) in multivariate and time series data using covariates. The goal is to develop machinery for covariance matrices just like the powerful generalized linear models (GLM) for the mean vector developed over two centuries. The key ideas and tools I rely on are from prediction theory, time series analysis and theory of stochastic processes. My interest in applications includes financial data analysis, analysis of longitudinal and panel data, data mining, classification and clustering, fMRI and high-dimensional data.
Bayesian statistics with a focus on spatial and spatio-temporal statistics
My research interests are in the fields of flexible regression methods for big data, and smoothing methods in nonparametric statistics. My current work focuses on choosing the shrinkage smoothing parameter for LASSO based on very large data sets.I also work with industry developing practical predictive models for big data in the fields of oil and gas, healthcare, hospitality, entertainment, financial services and retail.
Biostatistics, spatial statistics
Methodological research: missing data technique, measurement error, splines. Bayesian methods: parametric and nonparametric methods. Application: epidemiology, genetic epidemiology
Theory: Nonparametric smoothing methods, measurement error models, high dimensional inference and estimation, calibration and inverse problems.
Applications: in physical and engineering sciences: forensics, transportation, chemometrics, environmetrics, agriculture, and biology
Suhasini Subba Rao
Time series, nonstationary processes, nonlinear processes, recursive online algorithms, spatio-temporal models
Biostatistics, diffusions processes
Biostatistical inferences, missing and mis-measured data modeling and analysis, non- and semi-parametric methodology, resampling methods, small sample asymptotics, survey sampling
Stochastic models, directional data, mathematical statistics, nonparametric function estimation
Statistical methodology and application in bioinformatics, nutrition and epidemiology, functional/longitudinal data analysis
Empirical processes, bootstrapping