Improving the Performance of the k-Nearest Neighbors Algorithm with Utilization of the PSO Metaheuristic Algorithm

The k-nearest neighbor's algorithm (KNN) is one of the most widely used and useful nonparametric classification algorithms. The classification mechanism of this algorithm involves computing the distance between new instances and the instances whole classes are known. When the dataset contains non-numerical (ordinal and nominal) attributes, the performance of the algorithm can be significantly affected by how this distance is measured. In this paper, we attempt to improve the performance of the KNN algorithm by presenting a new solution for computing the distance of non-numerical traits. For this purpose, the Particle Swarm Optimization (PSO) algorithm is used. The task of this algorithm is to determine the best value of the distance between two states in a non-integer trait so that the accuracy of the KNN algorithm is increased. UCI University Learning Repository Data is used to test this idea. The results obtained from the proposed algorithm are compared with several other improved algorithms and show the useful improvement of this mechanism.