Bayesian Data Analysis for Biological and Behavioral Scientists
Dates
October 25, 2021 - Februrary 10, 2021
Instructors
Brendan Barrett and Urs Kalbitzer
Educational objectives
In this course for beginners and intermediates, participants will learn about the theoretical and practical foundations of Bayesian statistical inference using various types of linear models, causal inference, and the production of reproducible workflows.
Everyone is welcome (students, postdocs, professors), but only students should register to get credit. Please contact us if you would like to join, but are not registering.
Teaching Content
We will use the book Statistical Rethinking (McElreath 2020, 2nd edition) and the accompanying R package to learn and discuss the application of Bayesian data analysis. We will cover the basics of probability theory, statistical models and predictions, information criteria and model comparisons, sampling (Markov Chain
Monte Carlo), causal inference and DAGs, generalized linear models, mixture models, and introduction to hierarchical models.
Forms of Teaching
The course will consist of one seminar and one lab component each week. During the seminar part, we will discuss the content of one section of the book. During the lab part we will discuss and share solutions to the practical exercises described in the book.
Work load, examination, and unit completion
4 hrs/week attendance time and 6 hrs/week preparation time. Submission of written reports for the lab part each week.
Examination and unit completion
Submission of written reports for the lab part each week.
Prerequisites
All are welcome, but previous experience with linear models is beneficial and basic knowledge of R recommended. Inquire with instructors if you have any questions. Students should have a Github account set up and try to install R-Studio, rstan, and the rethinking package before first class.
Follow Up Course
We will offer a shorter subsequent course in Spring 2022 covering the remainder of the textbook: hierarchical models, measurement error, data imputation, Gaussian processes, and fitting theoretical models and mechanistic models to data. This will be geared towards experienced PhD Students and Postdocs.