Resolving The Conflicts Between Cuts In A Decision Tree With Verifying Cuts

A decision tree with verifying cuts, called V-tree, uses additional knowledge encoded in many attributes to classify new objects. The purpose of the verifying cuts is to confirm the correctness of the partitioning of tree nodes based on the (semi)-optimal cut determined by a greedy approach. The confirmation may be relevant because for some new objects there are discrepancies in the class prediction on the basis of the individual verifying cuts. In this paper we present a new method for resolving conflicts between cuts assigned to node. The method uses an additional local discretization classifier in each node where there is a conflict between the cuts. The paper includes the results of experiments performed on data sets from a biomedical database and machine learning repositories. In order to evaluate the presented method, we compared its performance with the classification results of a local discretization decision tree, well known from literature and called here C-tree, as well as a V-tree with previous simple conflict resolution method. Our new approach outperforms the C-tree, although it does not produce better results than V-tree with simple method of conflict resolution for the surveyed data sets. However, the proposed method is a step toward a deeper analysis of conflicts between rules.