A new technique for rule pruning in machine learning
Duc Truong Pham, Ziad Salem · 2004
This paper presents a simple but efficient post pruning method based on applying a decision tree induction algorithm to the rule set created by a rule induction algorithm. The proposed rule pruning method involves applying the ID3 decision tree induction algorithm to the set of rules produced by the RULES-4 covering algorithm. The result is a decision tree that can be converted into a more compact set of rules than the original rule set obtained with RULES-4. The results obtained using the new pruning method on a number of benchmark inductive learning problems will be presented to demonstrate its effectiveness.