Feature Subset Selection Based on Bayesian Networks

Bayesian network is a powerful tool of feature subset selection. Feature subset selection based on Bayesian network is to build the Markov blanket of class variable. In this paper, feature subset selection is done based on local dependency analysis method. First, basic dependency relationships between variables, basic structures between nodes, dependency separation criterion and the Markov blanket are analyzed. Then the Markov blanket of class variables is learned by dependency analysis. Finally, it is proved that learned feature subset is the Markov blanket of class variables under some assumptions. Experiments show that the method is more flexible, efficient and reliable than existing feature subset selection based on Bayesian network.