Model Similarity and Robustness in Predictions from Bayesian Networks

S.-H. Kim and G.Y. Noh (Korea)


Agreement level; Bayesian network; Model similarity; Conditional probability; Positive association.


Consider a prediction system based on a Bayesian network (BN) where all the variables involved are binary, each taking on 0 or 1. The system categorizes the probability that a certain variable is equal to 1 conditional on a set of variables in an ascending order of the probability values and predicts for the variable in terms of category levels. We introduce a similarity measure between BN models and describe how a BN model can be constructed which is similar to a given BN model. Then under the condition that all the variables are positively associated with each other, a method of obtaining an agreement level of predictions between two BN models is proposed. The agreement levels are obtained by a simulation experiment for a simple BN model.

Important Links:

Go Back