Enhancing Nearest Centroid with Coverage Principle for Classification Problem

Abstract

Motivated by the network coverage of the trans- mitter, this research proposes a novel coverage-based method to improve the Nearest Centroid’s class prediction by replacing the centroid with radius coverage as the reference in measuring distance. This novel approach, called Nearest Coverage, was tested using a breast cancer dataset to demonstrate its efficacy. The results indicate that this new classifier approach is promising more accuracy than the Nearest Centroid using appropriate coverage configurations. This method has the advantage of being straightforward and applicable to a wide variety of classification problems.