Creates a list containing the function name and arguments to train a OneClassSvm model with rxEnsemble.
oneClassSvm(cacheSize = 100, kernel = rbfKernel(), epsilon = 0.001, nu = 0.1, shrink = TRUE, ...)
The maximal size in MB of the cache that stores the training data. Increase this for large training sets. The default value is 100 MB.
A character string representing the kernel used for computing inner products. For more information, see maKernel. The following choices are available:
rbfKernel(): Radial basis function kernel. It's parameter represents
gammain the term
exp(-gamma|x-y|^2. If not specified, it defaults to
1divided by the number of features used. For example,
rbfKernel(gamma = .1). This is the default value.
linearKernel(): Linear kernel.
polynomialKernel(): Polynomial kernel with parameter names
degin the term
(a*<x,y> + bias)^deg. The
bias, defaults to
0. The degree,
deg, defaults to
ais not specified, it is set to
1divided by the number of features. For example,
maKernelPoynomial(bias = 0, deg = `` 3).
sigmoidKernel(): Sigmoid kernel with parameter names
coef0in the term
tanh(gamma*<x,y> + coef0).
gamma, defaults to to
1divided by the number of features. The parameter
0. For example,
sigmoidKernel(gamma = .1, coef0 = 0).
The threshold for optimizer convergence. If the improvement between iterations is less than the threshold, the algorithm stops and returns the current model. The value must be greater than or equal to
.Machine$double.eps. The default value is 0.001.
The trade-off between the fraction of outliers and the number of support vectors (represented by the Greek letter nu). Must be between 0 and 1, typically between 0.1 and 0.5. The default value is 0.1.
Uses the shrinking heuristic if
TRUE. In this case, some samples will be "shrunk" during the training procedure, which may speed up training. The default value is
Additional arguments to be passed directly to the Microsoft Compute Engine.