kernlab.svm.classif

kernlab.svm.classif-class(mlr)

R Documentation

Wrapped learner for Support Vector Machines from package kernlab for classification problems.

General hyperparameters:

C: Cost of constraints violation (default: 1), this is the 'C'-constant of the regularization term in the Lagrange formulation.

Kernel hyperparameters, sorted by kernel type:

rbfdot: sigma: inverse kernel width
laplacedot: sigma: inverse kernel width
polydot: degree: degree of the polynomial,
scale: scaling parameter of the polynomial,
offset: offset used in the polynomial
tanhdot: scale: scaling parameter of the tangent kernel,
offset: offset used in the hyperbolic tangent kernel
besseldot: sigma: inverse kernel width,
ordner: order of the Bessel function,
degree: degree of the Bessel function
anovadot: sigma: inverse kernel width,
degree: degree of the ANOVA kernel function
stringdot: length: length of the strings considered,
lambda: the decay factor,
normalized: logical parameter determining if the kernel evaluations should be normalized.

The kernel type and the hyperparameters are specified in parset.

initialize: signature( = "kernlab.svm.classif"): Constructor.
train.learner: signature(wrapped.learner = "kernlab.svm.classif", formula = "formula", data = "data.frame", weights = "numeric", parset = "list"): Overwritten, to allow direct passing of kernel hyperparameters. Besides that, simply delegates to super method.