The Class Estimation for Unseen Objects by Variable-Precision Dominance-based Rough Set Approach

Abstract

In this paper, estimation methods for decision attribute values are investigated based on the variable-precision dominance-based rough set model. The conceivable approaches are shown and the idea of k-NN is introduced to reduce the computation time. It is shown by numerical experiments that the proposed method together with k-NN is advantageous in both accuracy and computation time over the conventional estimation through rule induction.