Learning Bayesian networks structure based on prediction ability
Hui Wang, Shuangcheng Wang · Journal of Northeast Normal University · 2005
In the paper,the concept and estimating methods of the prediction ability are given.At the same time,it is proven that the prediction ability is the accurate of the prediction.The method of Bayesian networks structure learning based on prediction ability is developed.This method is made up of two parts:(1)setting up elementary Bayesian networks structure according to the absolute prediction ability;(2)regulating elementary Bayesian networks structure according to the conditional prediction ability and checking the loop.The experiment is made by simulation and the results are shown in the way of contrast,experimental results show that Bayesian networks structure can be learned by this method effectively.