Artificial neural networks, quantile regression, and linear regression for site index prediction in the presence of outliers

Authors

  • Carlos Alberto Araújo Júnior Universidade Federal de Minas Gerais, Instituto de Ciências Agrárias, Campus Regional de Montes Claros, Avenida Universitária, no 1.000, Bairro Universitário, CEP 39404-547 Montes Claros, MG.
  • Pábulo Diogo de Souza Universidade Federal de Santa Maria, Centro de Ciências Rurais, Departamento de Ciências Florestais, Avenida Roraima, no 1.000, Cidade Universitária, Camobi, CEP 97105-900 Santa Maria, RS.
  • Adriana Leandra de Assis Universidade Federal de Minas Gerais, Instituto de Ciências Agrárias, Campus Regional de Montes Claros, Avenida Universitária, no 1.000, Bairro Universitário, CEP 39404-547 Montes Claros, MG.
  • Christian Dias Cabacinha Universidade Federal de Minas Gerais, Instituto de Ciências Agrárias, Campus Regional de Montes Claros, Avenida Universitária, no 1.000, Bairro Universitário, CEP 39404-547 Montes Claros, MG.
  • Helio Garcia Leite Universidade Federal de Viçosa, Departamento de Engenharia Florestal, Avenida Purdue, s/no, Campus Universitário, Edifício Reinaldo de Jesus Araújo, CEP 36570-900 Viçosa, MG.
  • Carlos Pedro Boechat Soares Universidade Federal de Viçosa, Departamento de Engenharia Florestal, Avenida Purdue, s/no, Campus Universitário, Edifício Reinaldo de Jesus Araújo, CEP 36570-900 Viçosa, MG.
  • Antonilmar Araújo Lopes da Silva Celulose Nipo-Brasileira S.A., Rodovia MG 758, Km 3, s/no, Distrito Perpétuo Socorro, CEP 35196-000 Belo Oriente, MG.
  • Renato Vinícius Oliveira Castro Universidade Federal de São João Del-Rei, Departamento de Ciências Agrárias, Campus Sete Lagoas, Rua Sétimo Moreira Martins, no 188, Itapoã II, CEP 35702-031 Sete Lagoas, MG.

DOI:

https://doi.org/10.1590/S1678-3921.pab2019.v54.26491

Keywords:

Eucalyptus, artificial intelligence, dominant height, forest inventory, forest modelling, non-sampling errors

Abstract

The objective of this work was to compare methods of obtaining the site index for eucalyptus (Eucalyptus spp.) stands, as well as to evaluate their impact on the stability of this index in databases with and without outliers. Three methods were tested, using linear regression, quantile regression, and artificial neural network. Twenty-two permanent plots from a continuous forest inventory were used, measured in trees with ages from 23 to 83 months. The outliers were identified using a boxplot graphic. The artificial neural network showed better results than the linear and quantile regressions, both for dominant height and site index estimates. The stability obtained for the site index classification by the artificial neural network was also better than the one obtained by the other methods, regardless of the presence or the absence of outliers in the database. This shows that the artificial neural network is a solid modelling technique in the presence of outliers. When the cause of the presence of outliers in the database is not known, they can be kept in it if techniques as artificial neural networks or quantile regression are used.

Downloads

Published

2019-05-20

How to Cite

Araújo Júnior, C. A., Souza, P. D. de, Assis, A. L. de, Cabacinha, C. D., Leite, H. G., Soares, C. P. B., … Castro, R. V. O. (2019). Artificial neural networks, quantile regression, and linear regression for site index prediction in the presence of outliers. Pesquisa Agropecuaria Brasileira, 54(X), e00078. https://doi.org/10.1590/S1678-3921.pab2019.v54.26491

Issue

Section

QUANTITATIVE METHODS