How many components are there in the normal mixture model for the galaxy (Session 2)
Joint Psychology/Applied Statistics Colloquium
by Murray Aitkin
Abstract: Bayes analyses. Prior distributions and MCMC approaches to the posterior
distribution of model parameters.
How many components? - Posterior distributions for the number of
components. Conflicting conclusions.
A recent approach - the role of the likelihood ratio in simple model
comparisons. Nuisance parameter models - the Bayes factor and BIC.
Difficulties with improper priors.
Mapping the posterior distribution of the parameters to that of the
likelihood. Posterior distribution of the likelihood ratio. Computation
from the MCMC output.
How many components? - posterior distributions of the model likelihoods,
and those of the number of components.
For More Information: Owen Jones tel: 8344-6412 email: O.Jones@ms.unimelb.edu.au