Recent grants for Statistical research
Bayesian methods in support of risk management and asset pricing of large stock portfolios
Description: Several problems in finance revolve around the tails of a distribution and the uncertainty about its shape. To develop effective risk management solutions and asset pricing methods it is important to consider the uncertainty carried by the limited amount of information available. To achieve this, we propose to use an objective Bayesian approach to directly capture the tail uncertainty, by comparing tail behavior of hundreds of stocks to identify similarities and differences.
Empirical and bootstrap likelihood procedures for approximate Bayesian inference
2015 - 2017
Principal Investigator: Dr Fabrizio Leisen
Co-Investigators: Prof Brunero Liseo/Dr Clara Grazian (University of Rome La Sapienza)
Description: The overall scientific aim of this project is to provide new methods to analyse statistical models from a Bayesian perspective. In particular, we will focus on situations where the likelihood is intractable due to analytical or computational complexity. From the methodological point of view, we will tackle the problem with Empirical and Bootstrap Likelihood procedures. In particular, we will explore the new methodology in some specific challenging situations, where the orthodox Bayesian way is out of reach. The new findings will be tested on several datasets.
Flexible Bayesian non-parametric priors
2014 - 2018
Principal Investigator: Dr Fabrizio Leisen
Description: The use of Bayesian non-parametric (BNP) priors in applied statistical modeling has become increasingly popular in the last few years. From the seminal paper of Ferguson (1973, Annals of Statistics), the Dirichlet Process and its extensions have been increasingly used to address inferential problems in many fields. Examples range from variable selection in genetics to linguistics, psychology, human learning, image segmentation, and applications to the neurosciences. The aim of the project is to provide new flexible BNP priors for modelling different situations. In particular, two research lines will be investigated: vectors of BNP priors and non exchangeable species sampling sequences.
Advanced Bayesian Computation for Cross-Disciplinary Research
Description: Involving collaboration with Cambridge, Warwick, and Edinburgh universities, this project will develop new computational tools for Bayesian modelling that is flexible enough to capture complex phenomena and scalable enough to deal with very large data sets. One aspect of the project is the use of massively parallel graphics processing units (GPUs) to speed up computational modelling.
Statistical Ecology2017 - DAAD: Bilateral Exchange of Academics grant - Dr P T Besbeas
Statistical models for wildlife population assessment and conservation
2016 - 2017 and 2017 - 2019
Principal Investigator: Dr Rachel McCrea
Description: Within the environmental sector there is currently a shortage of practitioners equipped with the statistical modelling skills to carry out reliable population assessments. Consequently, environmental impact assessments (EIAs) and development mitigation projects often use population assessment protocols that are not fit-for-purpose.
The skills shortage arises because
- recent advances in statistical models for population assessment are largely confined to the academic sector with little penetration to the end-users; and
- although many postgraduate programmes have a statistical modelling training component, this often fails to expose PhD students to new models in the area and the potential applications these have for conservation practice.
This training programme will provide a cohort of PhD students and early career researchers/practitioners with the relevant modelling skills required for a career that involves wildlife population assessment for conservation.
Environmental modelling for moths and butterflies
Description: Long-term and large-scale datasets of species’ occurrence and abundance data for butterflies and moths provide valuable sources of information for monitoring and conserving these taxa. The work funded by this Fellowship focuses upon improving methods for modelling and analysis of UK butterfly and moth data.
New methods for the analysis of national butterfly and moth data
2015 - 2016
Principal Investigator: Prof Byron Morgan
Co-Investigator: Prof Martin Ridout
Funder: Butterfly Conservation
Description: Long-term and large-scale datasets of species’ occurrence and abundance data for butterflies and moths provide valuable sources of information for monitoring and conserving these taxa. This project, in collaboration with Butterfly Conservation, aims to improve methods for modelling and analysis of UK butterfly and moth data. This includes further development of novel methods for monitoring butterfly abundance, new analysis of occurrence data for moths, as well as work to quantify the outputs of citizen science data, such as from the Big Butterfly Count.
Integrated population modelling of dependent data structures
Description: The modelling of wild animal populations is of utmost importance in today's climate of global change. There is considerable threat to the survival of native species and it is necessary to determine why these threats are occurring and what can be done to prevent the loss of species forever. The mathematical modelling of animal populations facilitates the estimation of important demographic parameters and can confirm their relationship with spatial, environmental and individual covariates. Simple models were satisfactory for simple data sets. However, the development of sophisticated statistical models is severely lacking given the wealth of detailed individual level data being collected on a huge range of animal populations. This NERC-funded research fellowship is focussed on achieving the goal of developing an individual level model which accounts for fundamental correlations between data sets.
National Centre for Statistical Ecology - beyond 2010