A New Decision Tree Pruning Method Based on Rst

Pruning decision tree is an effective method to avoid the phenomena of overfitting. Various pruning methods have been proposed in many literatures. This paper gives a new decision tree pruning method based on Rough Set Theory (RST). According to the concept of explicit region in literature, this paper proposes two new concepts: depth-fitting ratio and error ratio to establish the new pruning strategy. Two thresholds c1 and c2 will be used to control the pruning extent. In addition, this paper gives a concrete example where it compares the new method with Pessimistic Error Pruning (PEP) which has proved the validity of the new one