MATH5960 is a Mathematics Level V course. See the course overview below.

Units of credit: 6

Prerequisites: there's no strict enforcement of prerequites for this course, however students are required to have a strong grasp of the principles of statistical inference. They are also expected to have prior experience with statistical computing using the software R or Splus.


Cycle of offering: every other year

Graduate attributes: the course will enhance your research, inquiry and analytical thinking abilities.

More information: this recent course handout (pdf) contains information about course objectives, assessment, course materials and the syllabus. (This pdf will usually be updated by the end of the first week of the semester.)

The Online Handbook entry contains up-to-date timetabling information.

If you are currently enrolled in MATH5960, you can log into the My eLearning Vista instance of this course.

Bayesian statistics could be described as the systematic application of probability to decision making in the face of uncertainty. It is a completely probabilistic approach to inference where we set up a full probability model for the data and unknowns in a problem and then condition on the data, making inference about unknowns from the conditional distribution of the unknowns given data (the so-called posterior distribution). Specification of a full probability model in a decision making problem involves specification of the likelihood function from classical inference but also specification of a prior distribution which expresses probabilistically what we know about the unknowns before observing data.

After describing the fundamentals of Bayesian inference this course will examine specification of prior distributions, links between Bayesian and classical frequentist inference, Bayesian model comparison and Bayesian computational methods. Markov chain Monte Carlo (MCMC) methods for computation will be described and implemented using the freeware statistical package WinBUGS. We will illustrate the advantages of the Bayesian approach by describing Bayesian inferential methods for a variety of models including linear models and various kinds of hierarchically structured models including mixture models.