A hierarchical clustering procedure that begins with each variable or observation in separate clusters. In subsequent steps, clusters that are closest together are combined to build a new aggregate cluster(s).