# How to calculate the Probability for the Unconditional Node in the Bayesian Belief Network?

In the popular example for Bayesian Belief network Burglary Alarm how is the probability for burglary P(B) and earthquake P(E) calculated as 0.001 and 0.002 respectively?

Is it an assumption made or there is some calculation involved ? I can understand the conditional probabilities for the child nodes but not sure how the probability for the nodes Burglary and Earthquake are getting calculated?

• Just looking at the slide deck. I think they are Priors (just known a priori). Assumptions, if you will. Just my 2 cents.
– knb
Mar 29, 2018 at 10:46

## 1 Answer

As the comment says, Burglary and Earthquake only have prior probabilities in the Bayes net. These priors are not calculated from other variables/distributions. They are just assumptions in this example, or, sugarcoated, expert knowledge. In fact, it is often hard to find good priors (from expert knowledge). Extracting them from large non-biased datasets can help. Anyway, one needs to understand the dataset and the context/circumstances/constraints of its collection.