Tree based frequent and rare pattern mining techniques: a comprehensive structural and empirical analysis

Abstract

Pattern mining has emerged as a compelling field of research over the years due to its extensive popularity in several application domains. Researchers in this field have contributed ample endeavors, encompassing different issues encountered during pattern generation. The patterns are either extracted based on a level-wise exploration of the search space or by employing efficient tree data structures. Tree based approaches show remarkable performance over the former ones on various grounds and are therefore regarded as the most prominent techniques of pattern mining. A precise and impartial analysis of these eminent techniques is necessary to widen the scope of effective pattern mining using the notion of tree data structures. This paper is therefore an attempt to provide a comparative scrutiny of the fundamental tree based pattern mining techniques through performance analysis based on several decisive parameters. The paper provides a structural classification of the widely referenced tree based pattern mining techniques in four categories and an analytical comparison of these techniques using benchmark real and synthetic datasets. Through this empirical study, an endeavor has been made to enable the researchers identify the factors affecting the performance of the most well-known techniques of pattern mining: the tree based approaches.