Classification and rule induction based on rough sets
J.W. Gryzmala-Busse, Chenchen Wang · Proceedings of IEEE 5th International Fuzzy Systems · 2002
Rules induced by machine learning systems from training data may be used for classification of new cases. The main objective of this paper is optimization of classification of unseen cases. In the experiments described in the paper, rules were induced by the system LERS (Learning from Examples based an Rough Sets). The classification system of LERS uses four parameters: strength-factor, specificity-factor, matching-factor and support. The paper shows the best choice of those four parameters in terms of error rate.