SVM with 'One-Versus-All' multiclass approach

Share:

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

1
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

A predict function which can be used to predict the classes for a new data set.

Author(s)

Patrick Warnat mailto:p.warnat@dkfz-heidelberg.de

See Also

MCRestimate

Examples

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
## 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)