A New Method for Improving the Performance of K Nearest Neighbor using Clustering Technique

In this paper, a new classification method is presented which uses clustering techniques to augment the performance of K-Nearest Neighbor algorithm. This new method is called Nearest Cluster approach, NC. In this algorithm the neighbor samples are automatically determined using clustering techniques. After partitioning the train set, the labels of cluster centers are determined. For specifying the class label of a new test sample, the class label of the nearest cluster prototype is used. Computationally, the NC method is faster than KNN, K times. Also the clustering techniques lead to find the best number of neighbors based on the nature of feature space. The proposed method is evaluated on two standard data sets, SAHeart and Monk. Experimental results show the excellent improvement both in accuracy and time complexity in comparison with the KNN method.