S.R. Gunn at the University of Southampton wrote Matlab SVM
Toolbox [27]. This toolbox runs in Matlab and is composed of m language
script files and functions, providing a good platformfor engineering
and practical application of the SVM technology. The regression
function of SVM is svr and is based on the training sample to design
the optimal regression function and to find the support vector. This
function has 6 parameters: input of the training sample, output of the
training sample, kernel function, penalty factor, loss function, and insensitivity
coefficient. The output parameter is the number of the support
vector, Lagrangian multiplier, and bias. Its grammar is