heuristicC: Fast Heuristics For The Estimation Of the C Constant Of A Support Vector Machine.

Description

heuristicC implements a heuristics proposed by Thorsten Joachims in order to make fast estimates of a convenient value for the C constant used by support vector machines. This implementation only works for linear support vector machines.

Usage

heuristicC(data)

Arguments

data

a nxp data matrix. Each row stands for an example (sample, point) and each column stands for a dimension (feature, variable)

Value

A value for the C constant is returned, computed as follows: \(\frac{1}{\frac{1}{n}\sum_{i=1}^{n}\sqrt{G[i,i]}}\) where \(G=\code{data}\%*\%t(\code{data})\)

References

T. Joachims SVM light (2002) http://svmlight.joachims.org

Examples

Run this code

# NOT RUN {
data(iris)

x=iris[,1:4]
y=factor(iris[,5])
train=sample(1:dim(iris)[1],100)

xTrain=x[train,]
xTest=x[-train,]
yTrain=y[train]
yTest=y[-train]

# Center and scale data
s=scale(xTrain,center=TRUE,scale=TRUE)

# Sparse Logistic Regression
t=6

co=heuristicC(s)
m=LiblineaR(data=s,labels=yTrain,type=t,cost=co,bias=TRUE,verbose=FALSE)

# Scale the test data
s2=scale(xTest,attr(s,"scaled:center"),attr(s,"scaled:scale"))

# Make prediction
p=predict(m,s2)

# Display confusion matrix
res=table(p$predictions,yTest)
print(res)

# Compute Balanced Classification Rate
BCR=mean(c(res[1,1]/sum(res[,1]),res[2,2]/sum(res[,2]),res[3,3]/sum(res[,3])))
print(BCR)

# }