A comparison of two partial matching strategies for classification of unseen cases

Abstract

This paper compares two partial matching strate- gies, selective and mixed, for classification of unseen cases. The selective partial matching is a novel approach for classification, while mixed partial matching was implemented in the LERS classification system several years ago. Though results of our experiments show that neither strategy is better than the other, an important conclusion is that it is crucial to implement both strategies since the correct choice of one of these strategies, for a specific data set, results in substantial improvement of the final classification.