Syllabus

Title
2035 Bayesian Computing
Instructors
Univ.Prof. Dr. Sylvia Frühwirth-Schnatter
Type
PI
Weekly hours
2
Language of instruction
Englisch
Registration
09/01/15 to 10/08/15
Registration via LPIS
Notes to the course
Dates
Day Date Time Room
Wednesday 10/21/15 05:00 PM - 08:00 PM D4.4.008
Wednesday 11/11/15 05:00 PM - 08:00 PM D4.4.008
Wednesday 11/18/15 05:00 PM - 08:00 PM D4.4.008
Wednesday 12/02/15 05:00 PM - 08:00 PM D4.4.008
Wednesday 12/16/15 05:00 PM - 08:00 PM D4.4.008
Wednesday 01/13/16 05:00 PM - 07:30 PM D4.4.008
Wednesday 01/20/16 05:00 PM - 07:30 PM D4.4.008
Wednesday 01/27/16 05:00 PM - 07:30 PM D4.4.008
Contents

This course is intended for doctoral and PhD students who want to gain a deeper understanding of Bayesian Computing. The following topics will be covered:

  • Introductory Bayesian computations (computing integrals, Monte Carlo methods, importance sampling)
  • Markov chain Monte Carlo methods (Metropolis-Hastings algorithm, Gibbs sampling, MCMC output analysis)
  • The principle of data augmentation (Gibbs sampling based on data augmention, simple methods for boosting MCMC)
  • Hierarchical models (random effects moels, mixture models, Bayesian regularization)
  • Bayesian model comparison and model selection (Bayes factors, marginal likelihoods, variable selection in a regression model, priors for model selection)
Learning outcomes

After completing this course the student will have the ability to:

  • Recall the basic principle of Bayesian inference (prior distribution, posterior distribution, predictive distribution, model evidence)
  • Recall the basic principle of Bayesian computing based on Monte Carlo simulation methods
  • Apply public domain packages for Bayesian inference and to analyse and evaluate the output of such packages
  • Design and implement computer programs for solving computational problems in Bayesian inference for commonly applied statistical models
Teaching/learning method(s)
This course is taught as a lecture combined with course projects which have to be solved individually by the students. In combination with the lecture, the course projects will help students to consolidate and expand their understanding of the theoretical and applied methods discussed in the lectures. For the course project, students create a report or a presentation of an R-based solution to a given computational task in Bayesian inference.
Assessment

Assessment will be solely based on the individual course projects submitted by each student. The assessment of the course projects will be based on the correctness of results, the clarity of the presentation, the ability to describe and apply the key concepts discussed throughout the course, and the recognizable effort made.

Readings
1 Author: Jim Albert
Title: Bayesian Computation with R

Publisher: Springer
Edition: 2nd Edition
Year: 2009
Recommendation: Strongly recommended (but no absolute necessity for purchase)
2 Author: Dani Gamerman and Hedibert F. Lopes
Title: Markov Chain Monte Carlo. Stochastic Simulation for Bayesian Inference.

Publisher: Chapman & Hall/CRC
Edition: 2nd Edition
Year: 2006
Recommendation: Strongly recommended (but no absolute necessity for purchase)
Availability of lecturer(s)
Per email: sylvia.fruehwirth-schnatter@wu.ac.at
Last edited: 2015-03-26



Back