4 day course: October 23 - October 26 2023
German Centre for Integrative Biodiversity Research (iDiv) Halle-Jena-Leipzig
benjamin.rosenbaum@idiv.de
The course offers a straightforward and practical approach to applied statistics using Bayesian inference. It starts with a gentle introduction to the concepts of Bayesian statistics (likelihood, priors, posterior distribution, MCMC sampling). Participants will both learn how to code models in the Stan environment and also get to know the user-friendly package ‘brms’. We will move step-by-step from basic linear regression to generalized, nonlinear, or mixed-effects models with a strong focus on the building blocks of statistical models.
Participants learn how to practically think in terms like data, model, likelihood, parameters, predictions. They learn how to specify and code statistical models of varying complexity in R. While the ‘brms’ package offers an easy transition from classical ‘lm’ or ‘lme4’ modeling, additional knowledge of ‘Stan’ code allows participants to adjust models to their specific research questions.
1 - Lecture: Introduction html
2 - Practical: Maximum likelihood html pdf
3 - Lecture: Bayesian principles html
4 - Lecture: Stan - a probabilistic programming language html
5 - Practical: my first Stan model html pdf
6 - Exercise: my first Stan model html pdf
1 - Lecture: Priors html
3 - Practical: Posterior distribution html pdf
4 - Exercise: Posterior html pdf
1 - Practical: Multiple regression (LM) html pdf
2 - Practical: Logistic regression (GLM) html pdf
3 - Practical: Nonlinear regression (NLM) html pdf
4 - Practical: T-test (LM) html pdf
1 - Practical: Fixed and random effects (LMM) html pdf
2 - Practical: Random intercepts regression html pdf
3 - Exercise: Random intercepts and slopes html pdf
4 - Lecture: Conclusions html