Author: Doug Leasure
This two-day course provides a practical and accessible introduction to Bayesian statistics for applied research in any field. Students will benefit from a combination of lectures and discussion to explore fundamental concepts unlocking the potential to design bespoke statistical analyses based on your data and hypotheses as well as practical exercises to gain hands-on experience implementing Bayesian models using free and open-source software. The course is designed as a springboard to overcome the steepest part of the Bayesian learning curve with an immersive two-day deep-dive.
Students can expect to gain a working understanding of Bayes theorem (i.e. likelihoods, priors, and posterior probability distributions) and implementation of Markov Chain Monte Carlo for Bayesian model fitting. The course will cover common probability distributions and the know-how to choose between them when designing bespoke models as well as a solid foundation in model validation techniques for high quality research. These concepts will be reinforced through practical lab exercises to gain experience implementing Bayesian models using R and Stan software. Students will be challenged through interactive group discussions to formalise their own mental models (i.e. hypotheses) into bespoke statistical models that can be used to confront these models with data.
Students will have the opportunity to schedule a follow-up Bayesian surgery appointment (30 mins in-person or remote) for one-to-one engagement (or small groups, as preferred) with the course instructor to answer burning questions that remain and/or to troubleshoot technical challenges related to their own research applications.
Pre-requisites: Students will need good programming skills in R and a basic understanding of linear regression to be successful in this course.
This repository is dual-licensed under MIT (code) and CC BY 4.0 (teaching materials). See the LICENSE file for details.