• JoomlaWorks Simple Image Rotator
  • JoomlaWorks Simple Image Rotator
  • JoomlaWorks Simple Image Rotator
  • JoomlaWorks Simple Image Rotator
  • JoomlaWorks Simple Image Rotator
  • JoomlaWorks Simple Image Rotator
  • JoomlaWorks Simple Image Rotator
  • JoomlaWorks Simple Image Rotator
  • JoomlaWorks Simple Image Rotator
  • JoomlaWorks Simple Image Rotator
 
  Bookmark and Share
 
 
Master's Dissertation
DOI
https://doi.org/10.11606/D.85.2018.tde-31102018-112210
Document
Author
Full name
Priscilla Ramos Carvalho
E-mail
Institute/School/College
Knowledge Area
Date of Defense
Published
São Paulo, 2018
Supervisor
Committee
Munita, Casimiro Jaime Alfredo Sepúlveda (President)
Barroso, Lucia Pereira
Ribeiro, Andreza Portella
Title in Portuguese
Estudo comparativo dos algoritmos hierárquicos de análise de agrupamentos em resultados experimentais
Keywords in Portuguese
análise de agrupamentos
Arqueometria
Coeficiente de Correlação Cofenética
INAA
métodos hierárquicos
Abstract in Portuguese
Objetivou-se, com este trabalho, estudar os métodos hierárquicos de análise de agrupamentos (ligação simples, ligação completa, ligação média, centróide e de Ward com base nas distâncias Euclidiana, Euclidiana ao quadrado, Manhattan e Mahalanobis), de modo a identificar qual é o mais adequado para uma base de dados arqueológicos. Utilizou-se uma base de dados fornecida pelo Grupo de Estudos Arqueométricos do IPEN CNEN/SP, na qual foram analisadas 146 amostras de fragmentos cerâmicos de três sítios arqueológicos por análise por ativação com nêutrons instrumental, sendo determinadas as frações de massa de 24 elementos químicos: As, Ba, Ce, Co, Cr, Cs, Eu, Fe, Hf, K, La, Lu, Na, Nd, Rb, Sb, Sc, Sm, Ta, Tb, Th, U, Yb e Zn. Para a determinação do melhor método, foram avaliados os dendrogramas conjuntamente com o valor dos coeficientes de correlação cofenética (CCC), obtidos para cada método. O método da ligação média mostrou-se mais coerente na formação dos agrupamentos, apresentando também os maiores valores do CCC. Por último, um script com funções do programa estatístico R foi desenvolvido para calcular o CCC, com o intuito de auxiliar os pesquisadores a encontrar o método de agrupamento mais apropriado para sua base de dados.
Title in English
Comparative study of hierarchical algorithms of cluster analysis in experimental results
Keywords in English
Archaeometry
cluster analysis
Cophenetic Correlation Coefficient
hierarchical methods
INAA
Abstract in English
The objective of this work was to study the hierarchical methods of cluster analysis (single linkage, complete linkage, average linkage, centroid and Ward with base in the Euclidean distance, Squared Euclidean distance, Manhattan and Mahalanobis) to identify which is the most appropriate in archaeological data set. It was used a data set provided by the Archaeometry Studies Group from IPEN CNEN/SP in which 146 samples of ceramic fragments from three archaeological sites analyzed by instrumental neutron activation analysis, being determined the mass fractions of 24 chemical elements: As, Ba, Ce, Co, Cr, Cs, Eu, Fe, Hf, K, La, Lu, Na, Nd, Rb, Sb, Sc, Sm, Ta, Tb, Th, U, Yb and Zn. For the determination of the best method, it was evaluated the dendrograms together with value of the cophenetic correlation coefficient (CCC) obtained for each method. The average linkage method was more coherent in formation of the clusters, and with higher CCC values. Finally, a script with functions of the R statistical software was developed to calculate the CCC with aim to be a tool for researchers find the most appropriate method of cluster analysis for their data set.
 
WARNING - Viewing this document is conditioned on your acceptance of the following terms of use:
This document is only for private use for research and teaching activities. Reproduction for commercial use is forbidden. This rights cover the whole data about this document as well as its contents. Any uses or copies of this document in whole or in part must include the author's name.
Publishing Date
2018-12-17
 
WARNING: Learn what derived works are clicking here.
All rights of the thesis/dissertation are from the authors
CeTI-SC/STI
Digital Library of Theses and Dissertations of USP. Copyright © 2001-2024. All rights reserved.