Statistics
|
Statistical Inference Research Group
The Group's research interests lie mainly in the provision of
fundamental statistical methodology for both formal and
computer-intensive statistical inference. Further details can be
found under Research interests
below.
We work closely with our colleagues in the Statistical Ecology
Research Group, whose activities are described on the web pages of
the
Research Unit for Wildlife Population Assessment and the Centre for Research into
Ecological and Environmental Modelling.
Staff and Research Students
Permanent staff
- Dr Ian B J Goudie
(mark-recapture, plant-capture, history of statistics)
- Dr Janine Illian
(spatial statistics, statistical methods for spatial and
spacio-temporal data especially spatial point process
modelling, functional data analysis, statistical modelling
of ecological communities especially for biodiversity)
- Dr Ruth King
(Bayesian inference, covariate data analyses, state-space
modelling)
- Dr Monique L
MacKenzie (generalized additive mixed models, splines)
- Dr Len Thomas
(sequential Monte Carlo methods, robust algorithms for
Bayesian and likelihood inference, state space models, model
selection)
- Dr Paul Wilson
(Statistical modelling, especially of zero-modified data;
Model Comparison methods)
Honorary staff
Current research students
Recent research students
Research interests
- Bayesian inference:
Bayesian inference involves distributions on the parameter
space.
Research topics in Bayesian inference in St Andrews include:
elicitation of prior knowledge, model selection and
model-averaging, efficient and generic reversible jump
Markov Chain Monte Carlo algorithms, application of
sequential Monte Carlo methods to estimation in state-space
models (mostly for wildlife population dynamics), random
effect models.
- Bioinformatics:
Real biological systems typically involve many components.
Typical current methods of data integration help
visualisation but not inference or prediction. Bayesian
integrated modelling of diverse data sources (e.g. genome
mapping, genetic, microarray, and proteomics data) is being
developed. Applications include cancer research, complex
phenotypes/diseases and comparative genomics.
- Differential geometry of parametric inference:
Asymptotic expansions can be expressed efficiently in the
language of differential geometry.
The invariant quantities which arise have geometric
interpretations which help to tame the complicated
expressions that occur when extending large-sample methods
for use with samples of moderate size.
- Directional statistics:
Observations which are directions, axes, or rotations (or,
more generally, points in a compact Riemannian manifold)
require special techniques.
- Estimation of population size:
Capture-recapture methods (especially coverage-adjusted and
conditionally-unbiased estimators), plant-capture (a variant
of mark-recapture under which a known number of planted
individuals is inserted into the target population),
distance sampling.
- Quantum statistical inference:
Measurements on quantum systems give rise to probability
distributions.
Inference on such distributions involves geometrical ideas
and issues of design.
- Reliability and survival analysis:
Models and inference are being developed for heterogeneous
data, in particular for data in which there are
(a) order restrictions on the dependence on covariates, (b)
dependence on duration.
- State-space modelling:
Research topics in state-space modelling in St Andrews
include: integrated data analyses, detection of
over-dispersion in the system process, use of sequential
Monte Carlo methods (such as particle filters) to fit
nonlinear state-space and related models to time series.
- Statistical genetics:
Association mapping methods help to identify genetic
elements responsible for a disease/phenotype, using data
from a general population.
Models and inference methods are being developed for
carrying out association mapping with fine-scale mapping,
heterogeneous populations, inbred populations, genome-wide
data, etc.
- Spatial statistics:
Spatial statistics is concerned with observations which have
location in the plane or in space.
Postgraduate Opportunities
We welcome enquiries from potential research students and
visitors who wish to work with the Statistical Inference
Research Group. Eligible British and European Union PhD students
can be funded (wholly or partially) through EPSRC Doctoral
Training Grants. Outstanding graduates of Scottish universities
are eligible for the prestigious Carnegie Trust.
Possible sources of funding for the overseas students not holding
grants from their home country are:
Applications to all the above grants are made through the School,
after the candidate has been offered a place. Suitably qualified
postgraduate students are offered the opportunity of supplementing
their income by undergraduate tutoring. For further information
about postgraduate study in Statistics at the University of St
Andrews, contact Dr
Madhuchhanda Bhattacharjee (Mathematical Institute,
University of St Andrews, St Andrews, Fife, KY16 9SS, Scotland).
Links to Further Information
Click here for information about the School of
Mathematics and Statistics, its Research, and
its Undergraduate
Teaching, and about the University of St Andrews
|