Statistical Model Selection

Statistical model selection investigates methods for choosing the best statistical model from a set of candidate models for a given dataset. This area addresses fundamental questions about how to define "best" and develops techniques to identify models that accurately represent underlying data patterns while avoiding overfitting. Research encompasses the development and application of criteria for model comparison, including information criteria, cross-validation, and hypothesis testing. Specific sub-fields include regression analysis, logistic regression, and the selection of models under budget constraints.

This research holds relevance for Arkansas by supporting data-driven decision-making across various sectors. In environmental science, it aids in selecting optimal models for water quality monitoring and understanding complex environmental chemistry. For economic development, it can inform budget-constrained model selection for resource allocation and policy analysis. The statistical methods developed also contribute to improving analytical techniques in fields like chemistry and chromatography, which have applications in Arkansas industries.

This work connects to analytical chemistry and chromatography, chemometrics, and environmental chemistry. Engagement spans multiple institutions within the state, fostering a collaborative environment for advancing statistical modeling techniques.