Explanatory Variables in Classifications and the Detection of the Optimum Number of Clusters
An ordinal approach to the a posteriori evaluation of the explanatory power of variables in classifications is proposed. The contribution of each variable is assessed in a way fully compatible with the distance or dissimilarity function used in the clustering process. Then, a simple ranking-based measure is applied to express the relative agreement or disagreement of variables with a given partition. This measure treats all variables equally, no matter how influential they were when the classification was actually created. The sum of measures for all variables reflects their overall agreement and can be used to select an optimal partition from a hierarchical classification.
KeywordsExplanatory Power Rank Order Hierarchical Classification Vegetational Plot Relative Agreement
Unable to display preview. Download preview PDF.
- Dale, M. B., Beatrice, M., Venanzoni, R. and Ferrari, C. (1986): A comparison of some methods of selecting species in vegetation analysis. Coenoses, 1, 35–52.Google Scholar
- Godehardt, E. (1990): Graphs as Structural Models: The Application of Graphs and Multigraphs in Cluster Analysis ( 2nd ed. ). Vieweg & Sohn, Braunschweig.Google Scholar
- Jancey, R. C. and Wells, T. C. (1987): Locality theory: the phenomenon and its significance. Coenoses, 2, 31–37.Google Scholar
- Lance, G. N. and Williams, W. T. (1977): Attribute contributions to a classification. Australian Computer Journal, 9, 128–129.Google Scholar
- Orldci. L. (1973): Ranking characters by a dispersion criterion. Nature, 244, 371–373.Google Scholar
- Orlóci, L. ( 1978 ): Multivariate Analysis in Vegetation Research. Junk, The Hague.Google Scholar
- Podani, J. (1985): Syntaxonomic congruence in a small-scale vegetation survey. Abstracta Botanica, 9, 99–128.Google Scholar
- Podani, J. (1994): Multivariate Data Analysis in Ecology and Systematics. SPB Publishing, The Hague.Google Scholar
- Ratkowsky, D. A. and Lance, G. N. (1978): A criterion for determining the number of groups in a classification. Australian Computer Journal, 10, 1 15–1 17.Google Scholar