The aim of classification is to obtain groups of objects (cases, variables) that are internally homogeneous and distinct from the other groups. When the variables (such as biological species) are classified, the homogeneity can be interpreted as their positive correlation, implying for species similar ecological behaviour, as reflected by the similarity of their distributions. The classification methods are usually categorised as in Figure 7–1.

Historically, numerical classifications were considered an objective alternative to subjective classifications, such as the classification of vegetation types by the Zürich–Montpellier phytosociological system (Mueller-Dombois & Ellenberg 1974; van der Maarel & Franklin 2013). It should be noted, however, that the results of numerical classifications are objective just in the sense that the same method gives the same results. Nevertheless, the results of all numerical classifications depend on the methodological choices, as we discuss in Section 7.3.1.

Example data set properties

The various possibilities of data classificationwill be demonstrated using vegetation data of 14 cases (‘relevés’) from Nízké Tatry Mts, already introduced in Section 6.5. Data were imported from the Excel file into a Canoco 5 project (TatryDCA.c5p). The primary data table was then exported into the condensed Cornell format (file tatry.dta) used by earlier versions of Canoco, to enable use of the TWINSPAN for Windows program. The data table present in the Excel file was also imported into the R software as a data frame called tatry, using the read.delim function.