APTS module: Statistical Modelling
Module leader: J J Forster and D. Woods
Previous module leaders: A C Davison and J J Forster (2007–8, 2008–9)
Please see the full Module Specifications 
document for background information relating to all of the APTS modules, including how to interpret the information below.
Aims: The main aim of this module is to introduce important general aspects of statistical modelling, including Bayesian modelling. A broad range of specific, commonly-used types of model will also be encountered.
Learning outcomes: After taking this module, students should — for topics listed below which are included in the module — understand the issues (why this is important), the terminology, the statistical principles associated with this aspect of modelling, and sufficient theory to deal with simple examples; and they will have gained some practical hands-on experience in more complex examples.
Prerequisites: Preparation for this module should (re-)establish familiarity with linear and generalized linear models, and with likelihood and Bayesian inference. Students who are familiar with (for example) chapters 4, 8, 10 and 11 of Davison's Statistical Models will be very well prepared (and will already know something of the areas to be covered in the module).
Topics:
- Missing data and latent variables.
- Principles and practice of model selection.
- Random-effects/hierarchical/mixed models.
- Semiparametric models and smoothing (links with the later APTS module 'Nonparametric Smoothing').
- The role of conditional independence in modelling. Introduction to graphical models.
Assessment: Either a suitably constructed 'comprehension exercise', for which students read a recent paper from the literature involving advanced modelling, and answer a series of questions, some of which may be quite open-ended; or a practical exercise involving real data and research questions.
