Data mining large abstract data sets for useful patterns is an attractive proposition. It may, for example, reveal useful trading rules in stock market data. There are two basic types of data mining, Automated data mining tools use algorithms that allow the computer to search the data. Perceptual data mining tools present the data to the user's senses (vision, hearing, touch) in a way that the user can search for useful patterns. The two methods are not disjoint, as rules discovered with the user's perception can then be automated. This paper describes a case study, where a visual-auditory interface was used to uncover patterns in stock market data. Results from a formal evaluation are reported. The paper also includes a discussion on how to incorporate these results into an automated tool using an agent framework.
|Title of host publication||ANZIIS 2003, Proceedings of the Eight Australian and New Zealand Intelligent Information Systems Conference|
|Editors||Brian Lovell, Duncan Campbell, Clinton Fookes, Anthony Maeder|
|Place of Publication||Brisbane, Australia|
|Publisher||Queensland University of Technology|
|Number of pages||6|
|Publication status||Published - 2003|
|Event||Australian and New Zealand Intelligent Information Systems Conference - Sydney, Australia, Australia|
Duration: 10 Dec 2003 → 12 Dec 2003
|Conference||Australian and New Zealand Intelligent Information Systems Conference|
|Period||10/12/03 → 12/12/03|
Nesbitt, K. (2003). Automated and Perceptual Data Mining of Stock Market Data. In B. Lovell, D. Campbell, C. Fookes, & A. Maeder (Eds.), ANZIIS 2003, Proceedings of the Eight Australian and New Zealand Intelligent Information Systems Conference (pp. 145-150). Queensland University of Technology.