Nalbantov, G.I.

error function model point value hyperplane vector support problem regression class variable method market kernel share figure support vector machines classication parameter distance machine training training data estimation brand nonlinear technique penalization data sets result algorithm loss function attraction estimate performance space solution object test point x market shares strategy number promotor optimization observation majorization margin example classification predictor variables panel instance-based approach prediction majorizing analysis month optimization problem svm-maj decision support hyperplanes vapnik complexity marketing dierent label accuracy huber erasmu period introduction output section university coecient table support vector regressions predictor classier