An assessment of numerical classificatory methods for geography.

Abstract

The development of the multivariate technique has grown parallel with the development and ready availability of digital computers. This paper attempts to assess the usefulness of one of these multivariate techniques, that is, the cluster analysis also commonly known as numerical classification. In this paper, procedures in developing clustering technique are described. As an illustration of this technique, continuous data of various sediments taken from 30 streams are used. Dissimilarity coefficient (the Euclidean distance) is used as measure of similarity and five fusion strategies are selected to produce various results.