EPSRC DTP PhD studentship: Estimation of risk profiles in spatio-temporal data

University of Exeter - College of Engineering, Mathematics and Physical Sciences

When the resolution of observed spatial and temporal data is higher than the effective scale of the underlying process, the choice of spatio-temporal unit can have a marked effect on the fitting and interpretation of statistical models, rendering some current algorithms infeasible.

The aim of this project is to explore Bayesian methods for optimally aggregating spatial/temporal units, where the level of aggregation is effectively treated as unknown and thus determined as part of the modelling. The project is strongly multidisciplinary, and potentially applicable to a wide range of real-world problems. There are also interesting statistical questions to be explored, for example, is the often observed overdispersion in spatio-temporal disease models partly attributable to analysing the data at an inappropriate level of aggregation?

Here, two main applications will be considered:

  • 1) The estimation of not only the degree of spatio-temporal risk in e.g. disease risk models, but also the optimal level of complexity and structure required in the spatial and temporal unit in order to efficiently estimate the parameters of interest. Areas of the space that have negligible background risks should be amalgamated to form larger regions, thus reducing the computational burden of the models.
  • 2) The estimation of the optimal temporal unit, for quantifying the effect of covariates on environmental processes. For example, determining the optimal way of defining “season” (i.e. the aggregation of days and months), in the study of extra-tropical cyclones (e.g. European storms) and the effect on their intensity and frequency from large-scale atmospheric processes such as the North Atlantic Oscillation. This particular problem is encountered in many weather and climate studies, where time is arbitrarily measured in days or months or seasons, units that mean different things depending on geographical location and the process of interest.

The student will attend the Academy for PhD Training in Statistics (APTS), in order to obtain a rounded view of modern statistical methods. They will be responsible for the development of novel statistical methods for spatio-temporal model fitting, and will be exposed to cutting edge statistical methodology. They will develop expertise in computer programming and numerical inference methods, and will learn key skills in science communication, and gain crucial experience in analysing complex, real-life data sets. These generic skills will serve the student well for a wide range of quantitative careers, as well as contributing scientific insights into a series of important real-world problems 

Funding Minimum
3.5 year studentship: UK/EU tuition fees and an annual maintenance allowance at current Research Council rate. Current rate of £14,296 per year.

Share this PhD
     
  Share by Email   Print this job   More sharing options
We value your feedback on the quality of our adverts. If you have a comment to make about the overall quality of this advert, or its categorisation then please send us your feedback
Advert information

Type / Role:

PhD

Location(s):

South West England