Computational Methods for Bayesian Statistics

The course is part of Mathematical Sciences, Master's Programme, University of Gothenburg, but is also open for students outside the program who meet the course prerequisites.

The course can be part of the following programmes: 1) Mathematical Sciences, Master's Programme (N2MAT) and 2) Applied Data Science Master's Programme (N2ADS).

The course requires a strong background in mathematics, at least one course in statistics, and skills in scientific programming (for example in R or Python) as achieved by completing TMS150 "Stochastic Data Processing and Simulation".

• Philosophy of Bayesian statistics.
• Conjugate priors and improper priors.
• Approximate methods for low-dimensional parameter spaces.
• Basic sampling methods.
• Monte Carlo integration.
• Advanced sampling methods, such as Markov chain Monte Carlo (MCMC).
• Hierarchical models.
• Computations for Bayesian Networks.
• Basic information theory.
• The EM algorithm.
• Basic variational Bayes methods.

On successful completion of the course the student shall be able to:

• explain and apply a Bayesian approach to probability inference,
• implement important computational algorithms for Bayesian inference, for example Metropolis-Hastings MCMC,
• make independent and informed decisions about statistical modeling and computational choices,
• present his or her analysis in a structured and pedagogical way.

Lectures and computer based hand-in assignments.

Computer-based hand-in assignments and follow-up workshops, with compulsory attendance and activity. The grade will be based on a written examination at the end of the course.

The grading scale comprises: Pass with Distinction (VG), Pass (G) and Fail (U).

Oral and/or written course evaluation will be performed. The results of the evaluation and possible changes to the course will be shared with students who participated in the evaluation and students who are starting the course.

The course Computational Methods for Bayesian Statistics (MSA102) has partially the same content as the courses Computational Methods for Bayesian Statistics (MSA101) and Computer Intensive Statistical Methods (MSA100). It is not allowed to be examined on more than one of these courses.