Estimation of the physical wood properties of Pinus taeda L. radial strips using least squares support vector machines. An Abstract from Journal of Near Infrared Spectroscopy

R.P. Cogdill,^a* L.R. Schimleck,^b P.D. Jones,^b G.F. Peter,^c R.F. Daniels^b and A. Clark, III^d
^aGraduate School of Pharmaceutical Sciences, Duquesne University, Pittsburgh, PA 15282, USA. E-mail: cogdillr@duq.edu
^bWarnell School of Forest Resources, The University of Georgia, Athens, GA 30602-2152, USA
^cSchool of Forest Resources and Conservation, University of Florida, Gainesville, FL, USA
^dUSDA Forest Service, Southern Research Station, Athens, GA, USA

Near infrared (NIR) spectroscopy offers a rapid method for estimating many important wood properties, including air-dry density, microfibril angle (MFA) and SilviScan estimated stiffness (EL(SS)). Wood property calibrations may be improved by using non-linear calibration methods. In this study, we compare calibrations developed using partial least squares (PLS) regression and least-squares support vector machine (LS-SVM) regression, a relatively new technique for modelling multivariate, non-linear systems. LS-SVM regression provided the strongest calibration statistics for all wood properties. For an equivalent number of latent variables, the predictive performance of the MFA LS-SVM calibrations were superior to those of the corresponding PLS calibration, while predictive results for air-dry density and EL_(SS) were similar for both calibration methods.