Modelling Spatial Processes
Statistical methods for spatial processes can be used for a variety of interesting applications in which stochastic events occur over space and time, such as modeling the incidence of disease in spatial epidemiology. However, posterior inference for such models is generally computationally very intensive, often taking days or even weeks to run a single Markov chain Monte Carlo based simulation. Our work aims to discover more efficient ways of characterising the complex probability distributions that result from these models, speeding up and increasing the accuracy of predictions in practice. This image shows examples of such 2 dimensional latent fields, processes, observed count data, and posterior inferences. The actual latent field and process are shown on top, with the data at the top-right and the posterior mean inferred fields, processes and associated variance from the model on the bottom.
For more information have a look at the following research paper:
- Girolami M. and Calderhead B. (2011) Riemann manifold Langevin and Hamiltonian Monte Carlo methods. J. R. Statist. Soc. B. (with discussion). 73, Part 2. pp 1-37.
This work is funded via the following research grants:
- Computational Statistics and Cognitive Neuroscience - EPSRC EP/H024875/1 - 2009 – 2011
- Inference-based Modelling in Population and Systems Biology - BBSRC BB/G006997/1 - 2010 to 2013
Page last modified on 21 jan 11 17:47
