DECISION TREE CLASSIFICATION USING ASSOCIATION MINING

Abstract

Association rule mining and Classification rule mining are two important data mining techniques. Association rule mining discovers a large set of rules that describe association relationships among a group of attribute values. These rules are rich and pass user specified minimum support and minimum confidence thresholds, but lack a systematic method of pruning overfitting rules for classification. Decision tree induction has an accuracy driven pruning, but imposes restrictive structures on rules. Both association and Decision tree classification. have a limitation in achieving the true classification. structure. This dissertation attempts to integrate association and classification mining techniques. The integration is done by focusing on mining a special subset of association rules whose consequent is restricted to the classification class attribute. The current work builds a decision tree like structure from these special association rules and applies the accuracy driven pruning of decision tree induction. The resulting classifier combines the richness of association rules and the accuracy driven pruning of decision tree induction. Experimental results show that the classifier built this way is, in most. of the cases, more accurate than that produced by the state-of-the-art classification system C4.5 and CBA. This dissertation work is implemented in Java programming language under windows environment.