A five-day intensive workshop covering probability, distributions, GLMs, mixed models, model selection, maximum likelihood, and Bayesian hierarchical modeling, using R and Nimble.
The Bayesian labs need a working toolchain, so install everything and confirm it runs before Day 1. Then download the course R project, which holds every dataset, script, and template in one folder per day: unzip it, double-click FW536.Rproj, and run 00_install_packages.R once. Learn the R Markdown workflow every hand-in uses. Then put whatever time you have left into the two R tutorials. Nearly every lab this week reshapes data with one or both of these toolkits, and time spent on them now is time you will not spend fighting syntax while the class moves on.
Each morning and afternoon has a single graded problem set, the best problems distilled into one coherent assignment, worth 20 points and submitted on Canvas.
Problems load real ecological data (salmon eDNA, Titanic survival, fish trophic position, ant richness, lizard occupancy, growth and recruitment data) so you practice the full workflow, not just formulas.
Explore distributions, link functions, shrinkage, and a live MCMC sampler. Drag sliders and watch the concepts move, right in the browser.
In-class practice labs have visible, reveal-able answers. The graded problem sets are separate and their keys are released only after grading.
Each day has a non-mathematical companion explaining the WHY before the HOW. Read it before lab to set up the concepts.
All problem sets are submitted as knitted R Markdown HTML. A tutorial covers install, knit, and embedding photos of written work.