FIT3229 · Bayesian modelling and inference

Overview

This unit introduces Bayesian approaches to statistical modelling, estimation, and prediction, with an emphasis on practical data analysis and computational methods. It develops the core ideas of prior distributions, posterior inference, Bayesian estimation, predictive distributions, credible intervals, and Bayesian approaches to hypothesis testing. You will apply these ideas in a range of modelling settings, including Bayesian regression models, latent variable models, Gaussian processes, kernel methods, and functional priors. The unit also introduces computational techniques used when exact Bayesian inference is not available, including Markov chain Monte Carlo, variational Bayes, and modern simulation-based approaches such as prior fitted networks. The focus is on developing conceptual understanding and practical modelling skills for students with a computing or data analysis background, rather than on advanced mathematical theory. By the end of the unit, you will be able to specify Bayesian models, interpret posterior and predictive uncertainty, apply approximate inference methods, and communicate Bayesian analyses clearly and responsibly.

Offerings

The Handbook publishes no offerings for this unit.

Assessment

The Handbook publishes no assessment items for this unit yet.

Assessment details may change. Please refer to the assessment information in Moodle closer to the start of the teaching period.

Requisites

Learning outcomes

1. Investigate complex problems by formulating Bayesian models, interpreting posterior and predictive uncertainty and evaluating alternative inference approaches with rigorous reasoning;
2. Select and apply contemporary computational tools including MCMC, variational Bayes and simulation-based methods to fit Bayesian models, attributing AI contributions transparently;
3. Implement, test and document Bayesian models and inference procedures in a high-level language, applying agreed conventions and verifying behaviour against realistic data;
4. Apply advanced Bayesian methods including regression, latent variable models, Gaussian processes and kernel methods to real data problems, interpreting and validating results;
5. Produce visualisations that communicate posterior distributions, credible intervals and predictive uncertainty to technical audiences;
6. Apply Bayesian machine learning methods to predictive tasks, evaluating model assumptions, prior choices and the implications for inference.

Workload

No workload detail published.

Student experience, not official rules. Nothing here changes what the Handbook says.