Clustering transactions with an unbalanced hierarchical product structure

Min Tzu Wang, Ping Yu Hsu, K. C. Lin, Shiuann Shuoh Chen

Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review

4 Scopus citations


The datasets extracted from large retail stores often contain sparse information composed of a huge number of items and transactions, with each transaction only containing a few items. These data render basket analysis with extremely low item support, customer clustering with large intra cluster distance and transaction classifications having huge classification trees. Although a similarity measure represented by counting the depth of the least common ancestor normalized by the depth of the concept tree lifts the limitation of binary equality, it produces counter intuitive results when the concept hierarchy is unbalanced since two items in deeper subtrees are very likely to have a higher similarity than two items in shallower subtrees. The research proposes to calculate the distance between two items by counting the edge traversal needed to link them in order to solve the issues. The method is straight forward yet achieves better performance with retail store data when concept hierarchy is unbalanced.

Original languageEnglish
Title of host publicationData Warehousing and Knowledge Discovery - 9th International Conference, DaWaK 2007, Proceedings
PublisherSpringer Verlag
Number of pages11
ISBN (Print)9783540745525
StatePublished - 2007
Event9th International Conference on Data Warehousing and Knowledge Discovery, DaWaK 2007 - Regensburg, Germany
Duration: 3 Sep 20077 Sep 2007

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume4654 LNCS
ISSN (Print)0302-9743
ISSN (Electronic)1611-3349


Conference9th International Conference on Data Warehousing and Knowledge Discovery, DaWaK 2007


  • Clustering
  • Data mining
  • Hierarchy
  • Similarity (distance) measure


Dive into the research topics of 'Clustering transactions with an unbalanced hierarchical product structure'. Together they form a unique fingerprint.

Cite this