MCRestimate (version 2.22.0)

SVM.OVA.wrap: SVM with 'One-Versus-All' multiclass approach

Description

Multiclass approach where k binary SVM classifiers are constructed for a classification problem with k classes: Every classifier is trained to distinguish samples of one class from samples of all other classes. For prediction of the class of a new sample, the sample is classified by all k classifiers, and the class corresponding to the classifier with the maximum decision value is chosen.

Usage

SVM.OVA.wrap(x,y,gamma = NULL, kernel = "radial", ...)

Arguments

x,y

x is a matrix where each row refers to a sample and each column refers to a gene; y is a factor which includes the class for each sample

gamma

parameter for support vector machines

kernel

parameter for support vector machines

...

Further parameters

Value

See Also

MCRestimate

Examples

## Not run: 
# library(golubEsets)
# data(Golub_Train)
# 
# class.column <- "ALL.AML"
# Preprocessingfunctions <- c("varSel.highest.var")
# list.of.poss.parameter <- list(var.numbers = c(250,1000))
# 
# Preprocessingfunctions <- c("identity")
# class.function <- "SVM.OVA.wrap"
# list.of.poss.parameter <- list(gamma = 6)
# plot.label <- "Samples"
# 
# cross.outer <- 10
# cross.repeat <- 20
# cross.inner <- 5
# 
# SVM.estimate <- MCRestimate(Golub_Train,
# 		class.column,
# 		classification.fun = class.function,
# 		thePreprocessingMethods = Preprocessingfunctions,
# 		poss.parameters = list.of.poss.parameter,
# 		cross.outer = cross.outer, cross.inner = cross.inner,
# 		cross.repeat = cross.repeat, plot.label = plot.label)
# 
# ## End(Not run)

Run the code above in your browser using DataCamp Workspace