African Journal of
Agricultural Research

  • Abbreviation: Afr. J. Agric. Res.
  • Language: English
  • ISSN: 1991-637X
  • DOI: 10.5897/AJAR
  • Start Year: 2006
  • Published Articles: 6843

Article in Press

Traditional regression versus Artificial Neural Networks models in sawing yield prediction

Ouorou Ganni Mariel Guera, José Antônio Aleixo Silva, Rinaldo Luiz Carraciolo Ferreira, Daniel Alberto Álvarez Lazo, Héctor Barrero Medel

  •  Received: 02 February 2019
  •  Accepted: 23 May 2019
The present work was developed with the objective of obtaining regression models and Artificial Neural Networks (ANNs) for accurate prediction of lumber recovery factor of Pinus caribaea var. caribaea in the sawmill Combate de Tenerías, belonging to Macurije forest company, Pinar del Río-Cuba. The inputs of the models were: log smaller diameter (D), diameter at breast height (DBH) and log taper or conicity (Con.). The lumber recovery factor prediction models were obtained, training ANNs and fitting regression models elaborated by means of Ordinary Differential Equations and stepwise regression. The best fit regression model was the model V, but its performance was lower than that of the ANNs MLP 3-3-1 and MLP 12-8-1, indicating that ANNs are viable techniques capable of providing equal or greater precision than those obtained with traditional regression models.

Keywords: Lumber, Ordinary Differential Equations, Algebraic Difference Approach (ADA), Artificial Neural Networks (ANNs), Multilayer Perceptron (MLP), Radial Basis Function (RBF).