EFFECTIVE ALGORITHMS FOR THE NEAREST-NEIGHBOR METHOD IN THE CLUSTERING PROBLEM

Two effective algorithms are presented for the nearest neighbor method in the hierarchical agglomerative clustering procedures. One is effective, when the number of clusters into which a data set should be classified is already known. The other is effective to search for several probable clustering solutions, when the number of clusters to be obtained is not known in advance. The computation times of the algorithms are shown to be O(N2) for clustering of N objects. Therefore, the algorithms are very powerful for the nearest neighbor method to classify a large data set.