I would like to find a simple but real example of
learning-the-values-of-a-Bayesian-network-with-known-structure.
I would appreciate if someone could indicate to me
a database containing:
- a list of realizations for a domain containing categorical
variables;
- a Bayesian network developed for that domain.
I have been able to find a lot of data in machine learning
repositories. Most of this data can be used to learn
*structure* and *values*. There are also many examples
of Bayesian networks where experts produced
*structure* and *values*. But I do not seem to find
an example where experts built the structure and them did
some statistical analysis to produce the values.
As far as I know, learning only values is a relatively
simple problem (at least it is simple compared to learning
values and structure). I have also seem arguments that
Bayesian networks have the advantage that structure can
be built first and values can then be estimated.
But I have not found easily something like: "this is the
causal structure for a domain X with categorical variables,
and this is real data obtained by measurements of variables
in domain X". Could someone help me find this?
Thanks much,
Fabio Cozman
Assistant Professor
University of Sao Paulo