Spectral decompositions in cluster analysis with applications to limnological data
Carregando...
Data
1995-03
Título da Revista
ISSN da Revista
Título de Volume
Editor
Associação Brasileira de Limnologia
Resumo
The establishment of groups based on dendrograms obtained through hierarchical cluster analysis is usually done in arbitrary way. In this paper, it is presented a method to detect at which level of clustering there is a maximization of the correlation between the calculated distance matrix among sampling unities (or operational taxonomic unities – OTUS’s) an a model matrix obtained from the dendrogram. The method is ilustrated by grouping sampling stations of six streams in the state of São Paulo (Brazil) by ten physical-chemical parameters of the water. The final classification obtained is correspondent to the expectation based on the ecological status of the system and its level of pollution.
Descrição
Palavras-chave
Análise do agrupamento, Classificação, Teste de mentel, Nível de corte, Cluster analysis, Classification, Mentel test, Cut level
Citação
BINI, Luis Mauricio; DINIZ FILHO, José Alexandre Felizola. Spectral decomposition in cluster analysis with applications to limnological data. Acta Limnologica Brasiliensia, Rio Claro, v. 7, n. 1, p. 35-40, mar. 1995.