There are several methods for inferring network structure in Bayesian networks, given data.

In my case I have a Bayesian network model built from old data, and I have a new source of data that I want to use to update the model, both in terms of structure and parameters. This new data source has a number of observations, much higher than the past data, making it ideal for structure inference. The problem is that the new data covers only a subset of the variables in the original model. Has anyone heard of a way to update BN model network structure (and parameters) given only a subset of the variables in the model?

  • $\begingroup$ Do you expect the new data to have bearing on the 'missing variables'? Idiot simple example: dataset with two variables: height and shoe size. A large new batch of height data should provides a smooth distribution of heights, which reliably describes the population. But will not change (i think) the relationship 'shoe size' ~ height. $\endgroup$ – Ivana Feb 3 '15 at 13:20

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.