Comparison Study on Forecasting of Timber Growth Ring Density with SVM and Neural Networks

Abstract

This paper made a comparison study on the forecasting of timber growth ring density with support vector machine (SVM) and radial basis function (RBF) neural network. The objective of this paper is to examine the feasibility of SVM in wood density forecasting by comparing it with a RBF neural network. Wood experiments are carried out to get the data sets. The simulation example shows that SVM outperforms the RBF neural network based on the criteria of normalized mean square error (NMSE), mean absolute error (MAE) and directional symmetry. Analysis of the experimental results proved that it is advantageous to apply SVMs to forecast wood density time series.