Bath Statistics Research

University of Bath

Research Areas

Environmental & Epidemiological Statistics

Gavin Shaddick is interested in all areas of statistical air pollution modelling, from modelling the pollutants themselves through to their effects on human health. Nicole Augustin is interested in environmental statistics, particularly in relation to European forest health and climate change.

Medical Statistics

Chris Jennison researches into the design and analysis of clinical trials. His book with Bruce Turnbull is a key text on the interim monitoring of clinical trials. His current research interests include: adaptive designs that allow treatment selection or enrichment of a subpopulation during a trial; adaptive designs for dose finding studies; optimisation of the combined Phase II / Phase III process.

Statistical Computing

Merrilee Hurn and Chris Jennison carry out research into innovative MCMC methods. Simon Wood is interested in reliable and efficient computational methods for semi-parametric and mixed models and is author of R recommended package mgcv for semi-parametric modelling and an accompanying textbook. Julian Faraway has written two very successful textbooks on using R for advanced regression modelling.

Smoothing and Functional Data

Simon Wood and Julian Faraway are interested in various aspects of smoothing and functional data analysis. Simon works on penalised regression spline smoothing and computational methods for estimating penalised GLMs. Julian works on functional data in the context of movement modelling.

Bayesian Methods

Bayesian analysis is a popular choice for complex problems with useful prior information and multiple sources of uncertainty. Merrilee Hurn and Chris Jennison have applied Bayesian methods to inverse problems arising in a variety of image analysis applications. Simon Shaw is interested in developing Bayes linear methods, in which expectation rather than probability is taken as primitive for expressing beliefs about random quantities of interest so that only a partial prior specification is required. Particular interests are in Bayes linear kinematics, for synthesising models with probabilistic and Bayes linear components and in applications to software testing. Merrilee Hurn works on Bayes estimates for image analysis and mixture models. The loss function plays a central role here, and she is interested in the case where the most reasonable loss does not admit closed form estimators. Evangelos Evangelou is working on prediction-optimal criteria for prior selection for the covariance parameters of latent Gaussian variables.

Information Geometry

Karim Anaya-Izquierdo is interested in the application of differential, convex, etc geometric methods to statistical inference.

Example publications

Anaya Izquierdo, K. A., Critchley, F. and Marriott, P. (2014) When are first order asymptotics adequate? A diagnostic. Stat 3(1) 17-22.

Augustin NH, Musio M, von Wilpert K, Kublin E, Wood SN and Schumacher, M (2009). Modelling spatio-temporal forest health monitoring data. Journal of the American Statistical Association, 104: 899-911.

Behrens G., Friel N. and Hurn M. (2012). Tuning tempered transitions Statistics and Computing, 22, 65-78.

Faraway J. and Choe S. B (2009) Modelling orientation trajectories Statistical Modeling 9 51-68.

Öhrn F. and Jennison C. (2010). Optimal group sequential designs for simultaneous testing of superiority and non-inferiority. Statistics in Medicine 29, 743–759.

Salway R, Lee D, Shaddick G and Walker S. (2010). Bayesian latent variable modelling in studies of air pollution and health . Statistics in Medicine, 29 (26): 2732-2742.

Wood S.N. (2010) Statistical inference for noisy nonlinear ecological dynamic systems. Nature 466(26): 1102-1104.

Some of our recent collaborators