A non-parametric method to estimate the number of clusters

Autor(es)

Fujita, Andre ^[1] ; Takahashi, Daniel Y. ^{[2, 3]} ; Patriota, Alexandre G. ^[4]

Afiliação do(s) autor(es)

^[1] Univ Sao Paulo, Inst Math & Stat, Dept Comp Sci, BR-05508 Sao Paulo - Brazil

^[2] Princeton Univ, Dept Psychol, Princeton, NJ 08544 - USA

^[3] Princeton Univ, Inst Neurosci, Princeton, NJ 08544 - USA

^[4] Univ Sao Paulo, Inst Math & Stat, Dept Stat, BR-05508 Sao Paulo - Brazil

• Tipo de documento: Artigo Científico
• Fonte: COMPUTATIONAL STATISTICS & DATA ANALYSIS; v. 73, p. 27-39, MAY 2014.
• Citações Web of Science: 24

Resumo

An important and yet unsolved problem in unsupervised data clustering is how to determine the number of clusters. The proposed slope statistic is a non-parametric and data driven approach for estimating the number of clusters in a dataset. This technique uses the output of any clustering algorithm and identifies the maximum number of groups that breaks down the structure of the dataset. Intensive Monte Carlo simulation studies show that the slope statistic outperforms (for the considered examples) some popular methods that have been proposed in the literature. Applications in graph clustering, in iris and breast cancer datasets are shown. (C) 2013 Elsevier B.V. All rights reserved. (AU)

• Processo FAPESP: 11/07762-8 - Causalidade de Granger entre grupos de séries temporais: desenvolvimento de metodologias para seleção de modelos e extensões no domínio da frequência com aplicações em biologia molecular e neurociência
• Beneficiário: André Fujita
• Modalidade de apoio: Auxílio à Pesquisa - Regular