Doctoral School

Bayesian Statistics

Abstract

Lately, Bayesian methods have gained popularity for tackling complex statistical tasks. This course covers Bayesian techniques from basic concepts to advanced topics like hierarchical modelling and model testing. Biostatistical examples illustrate theoretical concepts, combining classroom teaching with computer exercises. Practical learning is enhanced using JAGS and R. The curriculum is based on "Bayesian Biostatistics" by Lesaffre and Lawson (2012).

For specific information about the content, see 'Objectives' and 'Programme' below.

Target audience

PhD candidates who have good knowledge of regression techniques (linear, logistic, etc.) and some knowledge of models for correlated data such as mixed models. To be able to program in R is essential. 

Lecturer

Prof. Dr. Emmanuel Lesaffre is an emeritus professor at KU Leuven's Leuven Biostatistics and Statistical Bioinformatics Centre (L-BioStat). He has published over 300 articles and has been cited over 15,000 times, with an H-index of 58. Prof. Dr. Lesaffre specializes in biostatistics, longitudinal data analysis, and Bayesian methods, and is an experienced instructor in Bayesian statistics. He taught Bayesian statistics in the MatStat (Advanced Master in Statistical Analysis) program at UGent and at various other national and international institutions.

Organizing and scientific committee

Louise De Meulenaer (Department of Experimental Psychology, UGent)

Evelyne Fraats (Department of Experimental Psychology, UGent)

Objectives

  • Introduce Bayesian concepts and contrast them with the frequentist approach
  • Understand the background and the importance of computer-intensive methods in Bayesian statistical analyses, such as the Markov Chain Monte Carlo (MCMC) techniques: Gibbs and Metropolis-Hastings sampling;
  • Be able to work with Bayesian software, i.e. JAGS and its R interfaces.

Dates and venue

17, 18, 24 and 25 October 2024 (9h-18h)

Venue: On campus, to be announced

Programme

17 October

  • 9-13h: Frequentist statistics and the likelihood function and introduction to Bayes theorem and the posterior distribution-1

  • 14-18h: Introduction to Bayes theorem and the posterior distribution-2 and posterior summary measures and Bayesian hypothesis testing

18 October

  • 9-13h: Numerical techniques to determine the posterior distribution and introduction to Bayes theorem and the posterior distribution-1
  • 14-18h: Introduction to Bayes theorem and the posterior distribution-2 and more than one parameter

24 October

9-11h: Overview of the material seen on Days 1 & 2

11-13h: Introduction of Markov Chain Monte Carlo techniques

14-18h: Choosing the prior distribution and hierarchical models

25 October

9-11h: Model selection and model checking

11-13h: Working on group assignment

14-18h: Working on group assignment

Registration

Registration fee

Free of charge for PhD students and post-docs of a Flanders university. Otherwise, €250 for the whole course.

Number of participants

Minimum 10, Maximum 25

Language

English

Evaluation method

Criteria for evaluation are:

  1. 100% attendance
  2. Successful completion of a group assignment on the fourth day

After successful participation, the Doctoral School Office will add this course to your curriculum of the Doctoral Training Programme in Oasis. Please note that this can take up to one to two months after completion of the course.

More information

You can contact the organisers by e-mail.