Visually Contrast Two Collections of Frequent Patterns

• Published in: 2011 IEEE 11th International Conference on Data Mining Workshops pp. 1128 - 1135
• Main Authors: Carmichael, C. L., Hayduk, Y., Leung, C. K.
• Format: Conference Proceeding
• Language: English
• Published: IEEE 01.12.2011
• Subjects: association analysis; Contrast data mining; data and result visualization; Data mining; Data visualization; Educational institutions; human-machine interaction; Itemsets; Merchandise; pattern discovery; Time frequency analysis; Visualization
• ISBN: 1467300055; 9781467300056
• ISSN: 2375-9232; 2375-9259
• DOI: 10.1109/ICDMW.2011.177

Frequent pattern mining searches for frequently occurring sets of items or events. While users are interested in finding these frequent patterns in most situations, they may want to compare and contrast the mined frequent patterns in some other situations. For example, store managers may want to find out how the collections of frequently purchased items changed from one season to another. Similarly, regional managers may want to compare the frequently purchased items between two different branches. These are some examples of looking for temporal and/or spatial changes between mined frequent patterns. A visual representation of these patterns would be more comprehensive to users than the long textual list returned by many existing frequent pattern mining algorithms. However, many existing visualizers were not designed to show frequent patterns, let alone show the differences between them. In this paper, we propose a visualization system called Contrast Viz that enables users to visualize the mined frequent patterns and their differences.