Rare Association Rule Mining via Transaction Clustering

Rare association rule mining has received a great deal of attention in the recent past. In this research, we use transaction clustering as a pre-processing mechanism to generate rare association rules. The basic concept underlying transaction clustering stems from the concept of large items as defined by traditional association rule mining algorithms. We make use of an approach proposed by Koh & Pears (2008) to cluster transactions prior to mining for association rules. We show that pre-processing the dataset by clustering will enable each cluster to express their own associations without interference or contamination from other sub groupings that have different patterns of relationships. Our results show that the rare rules produced by each cluster are more informative than rules found from direct association rule mining on the unpartitioned dataset.