Neighborhood-Based Classification of Imprecise Data

Abstract

This chapter considers k-nn classification of objects that assume values of imprecise nature with respect to attributes being considered. In order to handle imprecise data values, two approaches of crisp conversion of fuzzy data sets are considered. The approaches considered in this chapter are borrowed from Credibility Theory of Liu. A comparative study on the choice of approach for crisp conversion of fuzzy data has been carried out with the help of certain multivariate simulated data sets. Conclusions drawn from the study are presented.