Hierarchical Statistical Modeling in Environmental Science
Hierarchical statistical models include generalized linear mixed models, generalized additive mixed models, state-space models for time-series data, random field models for spatial data, and specialized models such as occupancy, abundance, and demographic models for data from animal and plant populations. The course covers different aspects of these topics in different semesters, so students should look for an announcement from the instructor for an upcoming semester. Topics generally include formulation and analysis of such models with both frequentist and Bayesian methods, including Markov chain Monte Carlo. Students present and lead discussion on some topics. An additional unit may be earned for completing a course project.
Topics
- Students will be able to formulate and interpret the structure of a variety of hierarchical statistical models. They will understand and be able to interpret fixed and random effects, likelihood and its uses, model selection, different kinds of uncertainty quantification, Bayesian posterior sampling by MCMC algorithms, and the differences between frequentist and Bayesian methods. They will learn to use and interpret different estimation methods and results. Methods will include both frequentist and Bayesian methods, each of which have a variety of computational approaches. They will learn practical skills with some software tools. Specific learning outcomes will vary depending on the specific topics chosen for each iteration of the course. The additional learning outcome for students enrolled for three credits will be additional depth on course topics and experience from project-based data analysis and simulations.
Official catalogue course objectives and student learning outcomes
Sources
- Course catalogue ↗2026-2027