This four-day course introduces Bayesian inference and modelling. It is suitable for academics and professionals alike from diverse backgrounds ranging from industry to research fields such as population health, social sciences, disaster risk reduction, and many more. The focus will be on teaching the Stan interface, which works with statistical software packages to perform state-of-the-art statistical modelling within a Bayesian framework. The focus of this workshop will be on using Stan with R, although Stan can also be used with Python, Julia, Stata, and MATLAB. We will show you to develop and compile Stan scripts for Bayesian inference through RStudio to perform basic parameter estimation, as well as a wide range of regression-based techniques. This will begin with the simplest univariable linear models and its different families, moving up to Bayesian generalised linear models, hierarchical models and spatial intrinsic conditional autoregression models.
By the end of the workshop, the participants will:
- Understand the key principles of statistical modelling within a Bayesian framework;
- Be able to calculate inferential statistics on spatial and non-spatial data for hypothesis testing using the diverse types of regression-based models in a Bayesian framework
- Be able to perform spatial risk prediction for areal data as well as quantify levels of uncertainty using exceedance probabilities
- Acquire new programming language skills in Stan (interfaced with RStudio)
Each day will include a lecture, a live walkthrough demonstration of applying the methods, and a computer seminar class where participants will get hands-on experience of Bayesian estimation. For those that have not used R before, introductory materials to learn the basics of R programming will be provided before the course, making it suitable for users of all software packages.
Course instructor: Anwar Musah
Email address: a.musah@ucl.ac.uk