Condition Class Classification Stability in RST due to Continuous Value Discretisation
Rough Set Theory (RST) is a nascent technique for object classification, where each object in an information system is characterised and classified by a number of condition and decision attributes respectively. A level of continuous value discretisation (CVD) is often employed to reduce the possible large granularity of the information system. This paper considers the effect of CVD on the association between condition and decision classes in RST. Moreover, the stability of the classification of the objects in the condition classes is investigated. Novel measures are introduced to describe the association of objects (condition classes) to the different decision classes.
KeywordsCondition Class Decision Outcome Decision Class Classification Stability Component Probability
Unable to display preview. Download preview PDF.
- 2.Browne, C., Dünstch, L., Gediga, G.: IRIS revisited: A comparison of discriminant and enhanced rough set data analysis. In: Polkowski, L., Skowron, A. (eds.) Rough sets in knowledge discovery 2: Applications, case studies and software systems, pp. 345–368. Physica-Verlag, New York (1998)Google Scholar