Overview

hierarchical and non-hierarchical classification methods


Show Summary Details

Quick Reference

The grouping of individuals by a series of subdivisions or agglomerations to form a characteristic ‘family tree’ or dendrogram of groups. Alternatively, classification may be non-hierarchical, i.e. proceeding not by an organized series of progressive joinings or subdivisions, but instead achieving groupings by a series of simultaneous trial and error clusterings (successive approximation) until an optimum and stable pattern is found. A possible scheme, for example, is to select at random a number of starter individuals, equal to the number of groups required, and to which other individuals are added or from which they are removed, according to their characteristics and the classificatory philosophy used (that of seeking to minimize internal group heterogeneity, or of seeking to maximize differences between groups). The chief advantage of hierarchical over non-hierarchical methods is the clarity with which the routes to the final groupings may be followed, facilitating explanation of those groups. However, since most hierarchical classifications either dichotomize or pair at each division or join, natural clustering patterns may be distorted or poorly represented. Hierarchical classifications are also more likely to be unduly affected by irrelevant background information.

Subjects: Plant Sciences and Forestry.


Reference entries

Users without a subscription are not able to see the full content. Please, subscribe or login to access all content.