Dr Vee Ng

Room: S&C/3.006

Research

Statistical inference, change point problems.

Projects

Gibbs Samplers and Change-point Problems

The Gibbs sampler is a computationally intensive method of simulating multivariate random data via various conditional distributions which are often much easier to evaluate. It has found application in many areas of application and has made accessible problems which were previously regarded as intractable - particularly large-scale Bayesian problems. This project would review the use of Gibbs samplers in the Bayesian analysis of change-point problems, where for example regression slopes change at unknown points. The Gibbs sampler is used to obtain marginal posterior distributions of the changepoint and of the other model parameters, where the model involves a class of symmetric distributions with kurtosis measured by a non-normality parameter.

Back to Honours Projects