Bayesian Data Analysis DL (3792)
  
Coordinating lecturer :Prof. dr. Christel FAES 

Language of instruction : English
Credits: 3,0
  
Period: semester 1 (3sp)
  
2nd Chance Exam1: Yes
  
Final grade2: Numerical
 
Exam contract: not possible
 
Sequentiality
 
   Mandatory sequentiality bound on the level of programme components
 
 
  Following programme components must have been included in your study programme in a previous education period
    Introduction to Bayesian Inference DL (3579) 4.0 stptn
 

Prerequisites

The student has knowledge of basic concepts of Bayesian inference.


Content

Bayesian methods,statistical modeling

This course is the successor to the courseIntroduction to Bayesian Inference. Both courses are based on the bookBayesian Biostatisticsof Lesaffre and Lawson (2012, John Wiley & Sons). In the courseConcepts of Bayesian Inferencesome basic Bayesian principles have been introduced anditwas taught how to use WinBUGS/OpenBUGS/Jags/Nimble. These skills will be assumed in the present course.NewBayesian topics will be introduced but also thetopics introduced in the previous course will be further developed. An example of the latter is that wewill explorethe role of the prior distribution in more detail, but also that the MCMC techniques will be explained in more technical detail. In this course,Bayesian methods for model selection and model criticism will be treated. In addition,more complex statistical analyses will be tackled e.g. Bayesian approaches for missing data, smoothing, survival analysis, etc.

Contents

Exploring the prior distribution

  • The conjugate prior distribution: derivation, further examples and semi-conjugacy
  • Expressing ignorance: the noninformative prior, the vague prior, Jeffreys prior, improper priors
  • Informative priors: data-based priors, elicitation of prior knowledge, archetypal priors, prior distributions of regression models, modeling priors

Alternatives to Markov Chain Monte Carlo techniques

  • INLA
  • Hamiltonian Monte Carlo Methods

Hierarchical models

  • Poisson-gamma hierarchical model
  • Gaussian hierarchical model
  • Comparison full Bayesian approach with empirical Bayesian approach
  • Bayesian mixed models: linear, generalized linear and non-linear models
  • Miscellaneous topics: choice of level 2 variance prior, propriety of posterior, comparison with frequentist approaches

Model building and assessment

  • Bayes factor and variants, e.g. pseudo-Bayes factor
  • Model selection based on predictive loss functions
  • Residual analysis
  • Sensitivity analysis
  • Posterior predictive checks
  • Model expansion techniques


Organisational and teaching methods
Organisational methods  
Collective feedback moment ✔  
Distance learning ✔  
Project ✔  


Evaluation

Semester 1 (3,00sp)

Evaluation method
Other evaluation method during teaching period50 %
Other Group project
Transfer of partial marks within the academic year✔
Oral exam50 %
Open questions✔
Use of study material during evaluation✔
Explanation (English)Slides and notes
Additional information The maximum score on the group project is 10. The maximum score on the individual oral exam is 10. When the student scores less than 5 on the individual oral exam, then one point will be subtracted from the global (group project + individual oral exam) score of the course. When the student scores less than 2,5 on the individual oral exam, then two points will be subtracted from the global score.

Second examination period

Evaluation second examination opportunity different from first examination opprt
No
 

Previously purchased compulsory textbooks
  Bayesian Biostatistics,E. Lesaffre and A. Lawson,1st,John Wiley & Sons, 2012,9780470018231,for course "Introduction to Bayesian Inference"
 

Compulsory course material
 

Medical papers on the use of Bayesian methods in clinical trials and epidemiology. For this one needs to consult the scientific literature

Course notes "Bayesian Data Analysis II".

Lesaffre, E. and Lawson, A. Bayesian Biostatistics, John Wiley & Sons, 2012 The course notes are strongly linked to the book and further explanations of topics treated during class can be found in the book.

The R software will be used in this course.



Learning outcomes
Master of Statistics and Data Science
  •  EC 

    • The student can handle scientific quantitative research questions, independently, effectively, creatively, and correctly using state-of-the-art design and analysis methodology and software.

     
  •  DC 

    • ... correctly using state-of-the-art analysis methodology.

     
  •  DC 

    • ... correctly using state-of-the-art software.

  •  EC 

    • The student can critically appraise methodology and challenge proposals for and reported results of data analysis.

  •  EC 

    • The student can work in a multidisciplinary, intercultural, and international team.

  •  EC 

    • The student is an effective written and oral communicator, both within their own field as well as across disciplines.

     
  •  DC 

    • The student is an effective writer in their own field.

  •  EC 

    • The student is capable of acquiring new knowledge.

 
  EC = learning outcomes      DC = partial outcomes      BC = evaluation criteria  
Offered inTolerance3
second year Data Science - distance learning J
second year Master Bioinformatics - distance learning J
second year Master Biostatistics - distance learning J
second year Quantitative Epidemiology - distance learning J



1   Education, Examination and Legal Position Regulations art.12.2, section 2.
2   Education, Examination and Legal Position Regulations art.15.1, section 3.
3   Education, Examination and Legal Position Regulations art.16.9, section 2.