Research Projects

Long-Term Methodological Projects:

The characteristics of available data often influence application driven methodological research. Thus, the advent of remote sensing instruments, high-throughput technology, social networking, and other technologies, has led to a new era in statistics. Specifically, methodology is needed for analyzing high-dimensional data that have many different sources of dependency. There are natural difficulties that occur when analyzing these modern datasets. In particular, one would like to leverage information from these dependencies to improve the precision of estimates. However, this may not be possible due to the dimension of the dataset. My research focus is primarily devoted to developing application driven methodology for big spatio-temporal data that exhibit dependencies over different variables and/or spatio-temporal scales. I often refer to this type of data as "complex dependent data," and by "complex" I mean that there are several sources of dependency.

To develop models that allow for several sources of dependency it is often easier to start with fewer sources of dependency, and as a result, much of my work has been done in the spatial-only data setting. Also, several of the methods I propose are Bayesian in nature, and consequently, I have been developing some Bayesian methods with the purpose of analyzing complex dependent data.

The links below lead to more details on my research.

• Complex Dependent Data

  • Multivariate Spatio-Temporal Data

  • Multiscale Spatio-Temporal Data

• Spatial Statistics

• Bayesian Analysis

  • Distribution Theory for Dependent Data

  • Bayesian Model Building for Dependent Data (In preparation)

Applications of Primary Interest:

Multivariate spatio-temporal and multiscale spatio-temporal datasets have become commonplace, and there is a clear need to develop statistical methodology to analyze them. For example, multiscale multivariate spatio-temporal public-use estimates are provided through the US Census Bureau’s American Community Survey (ACS). In particular, there are 1-year and 5-year multiyear period estimates (MYE), and margins of errors, for multiple demographic and socio-economic variables over different predefined geographies. Remote sensing instruments collect a countless number of variables over the globe and over discrete time. The CDC and NIH collect several health indicators over US counties over time, in-part, to monitor the well-being of US citizens. The nature of these datasets have motivated me to answer important inferential questions to official statistics, environmental statistics, and survival analysis.

• Official Statistics

• Environmental Statistics

• Survival Analysis