Computing Optimized Rectilinear Regions for Association Rules

We address the problem of nding useful regions for two-dimensional association rules and decision trees. In a previous paper we presented ecient algorithms for computing optimized x-monotone regions, whose intersections with any vertical line are always undivided. In practice, however, the quality of x-monotone regions is not ideal, because the boundary of an xmonotone region tends to be notchy, and the region is likely to overt a training dataset too much to give a good prediction for an unseen test dataset. In this paper we instead propose the use of a rectilinear region whose intersection with any vertical line and whose intersection with any horizontal line are both undivided, so that the boundary of any rectilinear region is never notchy. This property is studied from a theoretical viewpoint. Experimental tests conrm that the rectilinear region less overts a training database and thefore provides a better prediction for unseen test data. We also present a novel ecient algor...

• Type: Conference
• Year: 1997
• Where: KDD
• Authors: Kunikazu Yoda, Takeshi Fukuda, Yasuhiko Morimoto, Shinichi Morishita, Takeshi Tokuyama
• Keywords: Data Mining, KDD 1997, Rectilinear Regions, Vertical Line, X-monotone Regions