next up previous
Next: Conclusions Up: Related Work Previous: Subgroup Evaluation Measures

5.4 Subgroup Visualization

Data visualization methods have been part of statistics and data analysis research for many years. This research concentrated primarily on plotting one or more independent variables against a dependent variable in support of exploratory data analysis (Tukey, 1977; Lee, Ong, & Quek, 1995; Unwin, 2000).

The visualization of analysis results has, however, gained only recently some attention with the proliferation of data mining (Card, Mackinlay, & Shneidermann, 1999; Fayyad, Grinstein, & Wierse, 2002; Keim & Kriegel, 1996; Simoff, Noirhomme-Fraiture, & Boehlen, 2001). The visualization of analysis results primarily serves four purposes: better illustrate the pattern to the end user, enable the comparison of patterns, increase pattern acceptance, and enable pattern editing and support for ``what-if questions''. The recent interest in the visualization of analysis results was spawned by the often overwhelming number and complexity of data mining results.

Readers interested in comparing the visualization method proposed in this paper with other subgroups visualization methods can find the visualization of subgroups A1-C1 in the joint work by Gamberger, Lavrac, & Wettschereck (2002).