Analyzing Attribute Dependencies

Aleks Jakulin and Ivan Bratko (2003) Analyzing Attribute Dependencies. In: 7th European Conference on Principles and Practice of Knowledge Discovery in Databases (PKDD 2003), September 22-26, 2003, Cavtat, Croatia.

There is a more recent version of this item available.

Preview

PDF Download (138Kb)

Abstract

Many effective and efficient learning algorithms assume independence of attributes. They often perform well even in domains where this assumption is not really true. However, they may fail badly when the degree of attribute dependencies becomes critical. In this paper, we examine methods for detecting deviations from independence. These dependencies give rise to “interactions” between attributes which affect the performance of learning algorithms. We first formally define the degree of interaction between attributes through the deviation of the best possible “voting” classifier from the true relation between the class and the attributes in a domain. Then we propose a practical heuristic for detecting attribute interactions, called interaction gain. We experimentally investigate the suitability of interaction gain for handling attribute interactions in machine learning. We also propose visualization methods for graphical exploration of interactions in a domain.

Available Versions of this Item.

• Analyzing Attribute Dependencies (deposited 22 Jan 2004)[Currently Displayed]
  • Analyzing Attribute Dependencies (deposited 04 Apr 2005)

Actions (login required)