KTDA: emerging patterns based data analysis system

Abstract

Emerging patterns are kind of relationships discovered in databases containing a decision attribute. They represent contrast characteristics of individual decision classes. This form of knowledge can be useful for experts and has been successfully employed in a field of classification. In this paper we present the KTDA system. It enables discovering emerging patterns and applies them to classification purposes. The system has capabilities of identifying improper data by making use of data credibility analysis, a new approach to assessment data typicality.