Variable Precision Rough Sets in Analysis of Inconsistent Decision Tables
Alicja Mieszkowicz-Rolka, Leszek Rolka · 2003
This paper discusses the idea of variable precision rough sets, which can be viewed as an extension of the basic rough set notion. A new definition of the positive area of classification is proposed and investigated. This approach leads to a change of properties of the approximation quality, which then becomes a more suitable measure for assessment of dependencies between condition and decision attributes in decision tables obtained from real processes. It was shown that the approximation quality can be equally obtained by using the notion of filling error of a set. The conclusions of the paper concern the admissible inclusion error a and present the comparison of the investigated measures. The propositions can be effectively applied to analysis of information systems in case of dynamic processes controlled by a human operator.