A new approach to isotonic agglomerative hierarchical clustering (Q1203108)

A clustering is a type of classification imposed on a finite set of objects. In the paper a new approach to agglomerative hierarchical clustering is presented. The new agglomerative algorithms presented are described by a level statistic to determine the level of joining the clusters and the selection statistic which selects the pairs of clusters to be joined. For traditional agglomerative algorithms, the level statistic and selection statistic are identical. Two examples concerning generalizations of complete linkage and average linkage are illustrated.