1
$\begingroup$

I am using bnlearn package and R to learn Bayesian Network structure and also fit it using Maximum Likelihood estimation(MLE). bn.fit uses MLE to learn[as i understand] a generalized probability distribution for all data. However, it is required to obtain probability distribution for a particular variables given observed (evidence) data from other variables. There one straighforward answer is to sample required variable for given evidence variables and fit probability distribution from that data. I am new to bnlearn package and Bayes Nets and thinking of maybe there is more natural way of obtatining those probability distribution parameters?

Improve this question
$\endgroup$
1
  • $\begingroup$ (+1) Please elaborate on some points to make your question more clear: a) generalized probability distribution for all data. b) ...for a particular variables given observed (evidence) data from other variables. c) ...required variable for given evidence variables and fit .... $\endgroup$
    – naive
    Commented Feb 27, 2019 at 7:07

0

Reset to default

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.