Fitting hierarchical models in Template Model Builder

Course materials for a virtual TMB workshop at the Universidad de Concepción, January 2022.

Tenative syllabus:

	Topic	Morning Topic (Theory and Mechanisms) 1.5 hour 10am-11:30 CST 5am-6:30am PST Definitely recorded	Late Morning - Case-Studies / Projects 1 hour [C&N] 12-1pm CST 7-8am PST Potentially recorded	Afternoon Lab 3 hours [C&N] 2-5pm CST 9-12pm PST Live –virtual
Monday	Course introduction; Review model fitting; introduce TMB	Numerical optimization using derivatives; Finite differences vs autodiff; Refresher on Rstudio & R. [Cole]	Maximum likelihood of single parameter Normal model (Excel and R) [Cole]	Poisson model Project time: proposals and scope
Tuesday	Building linear models in TMB	TMB/C++ syntax, workflow, factors, debugging, factors in TMB [Cole]	Maximum likelihood estimation; Fit LM in R by hand, using lm() and in TMB [Cole]	Exploring Beverton-Holt TMB model
Wednesday	GLMs	GLM theory [To do] [Noble]	Estimating uncertainty: standard errors, Delta method [Noble] likelihood profile, bootstrapping	Poisson GLM Project time
Thursday	Random effects in TMB	Types of mixed effects models; Marginal maximum likelihood [Noble]	Project time	VBG for pooled samples and individual ones; Hierarchical VBG;
Friday	Beyond GLMMs	Advanced TMB features: mapping, REML, simulation, functions sdmTMB and glmmTMB? [To do] [Cole?]	Project time	Project presentations

Some material was modified with permission from a course by Jim Thorson at https://github.com/James-Thorson/2016_Spatio-temporal_models.

Ideas for next time

Data sets

• Just have one data set that we keep analyzing over and over
  • Has fixed effects
  • Has random effects
  • Has a slope
• First day - Start with just a mean and normal likelihood
• Second day - add the slope
• Third day - add random effects and fit with LME
• Fourth day - add GLM

Advantages of recording

• Was good for presenting ideas and could be used in the future

Uses of channels

• Could create multiple rooms for groups of 2
• Use them in the labs so that 2 students can share a room
• Professores can move among rooms and check in on the students working on the labs and projects

Topics on which to focus an entire class

• CPUE analysis
• glmmTMB and sdmTMB?