Bayesian data analysis is the art of extracting information from
data using Bayesian methods. There are three main aspects of data
- model building
- model refinement
- model validation
Model building: I use
graphical models and Bayesian networks for modeling complex dependency
Model refinement: This
is the task of refining a class of models using data. My expertize
is in developing methods for automating this process. Methods I
developed for automated learning with incomplete data are implemented
in Discoverer and RoC.
I am currently developing a greedy search algortihm for inducing
Bayesian networks from continuous data.
Model validation: This
is the task of validating a model refined from data. My current
interest is in developing diagnostic tools to assess the goodness
of Bayesian networks learned from data.