Rosabeth White
Project title
'Composite score summary statistics for approximate Bayesian computation'
Project Description
Approximate Bayesian computation (ABC) and composite likelihoods can be useful tools if the full likelihood is unavailable. A composite likelihood can be used as a replacement to the full likelihood. ABC methods avoid direct evaluation of the likelihood by instead comparing summary statistics of simulated data to the observed data. The quality of an ABC approximation often depends heavily on the choice of summary statistics. My current research aims to develop new summary statistics for ABC using composite likelihoods.
Research Interests
- Computational statistics
- Bayesian inference
- Likelihood-free methods
Education
I graduated in 2018 with an MMathStat from Newcastle University.
Supervisors
Contact
Email: r.white4@ncl.ac.uk