vcr.svm.train {classmap}

Prepare for visualization of a support vector machine classification on training data.

Description

Produces output for the purpose of constructing graphical displays such as the classmap. The user first needs to run a support vector machine classification on the data by e1071::svm, with the option probability = TRUE. This classification can be with two or more classes. The output of e1071::svm is then an argument to vcr.svm.train. As e1071::svm does not output the data itself, it needs to be given as well, in the arguments X and y.

Usage

vcr.svm.train(X, y, svfit, ortho = FALSE)

Arguments

X

matrix of data coordinates, as used in e1071::svm. Missing values are not allowed.

y

factor with the given (observed) class labels. It is crucial that X and y are exactly the same as in the call to e1071::svm.

svfit

an object returned by e1071::svm, called with exactly the same X and y as above.

ortho

If TRUE, will compute farness in the orthogonal complement of the vector beta given by e1071::svm. Is only possible for 2 classes, else there would be several beta vectors.

Value

A list with components:

yint

number of the given class of each case. Can contain NA's.

y

given class label of each case. Can contain NA's.

levels

levels of the response y.

predint

predicted class number of each case. Always exists.

pred

predicted label of each case.

altint

number of the alternative class. Among the classes different from the given class, it is the one with the highest posterior probability. Is NA for cases whose y is missing.

altlab

label of the alternative class. Is NA for cases whose y is missing.

PAC

probability of the alternative class. Is NA for cases whose y is missing.

figparams

parameters used in fig, can be used for new data.

fig

distance of each case i from each class g. Always exists.

farness

farness of each case from its given class. Is NA for cases whose y is missing.

ofarness

for each case i, its lowest fig[i,g] to any class g. Always exists.

svfit

as it was input, will be useful for new data.

X

the matrix of data coordinates from the arguments. This is useful for classifying new data.

Author(s)

Raymaekers J., Rousseeuw P.J.

References

Raymaekers J., Rousseeuw P.J., Hubert M. (2021). Class maps for visualizing classification results. Technometrics, appeared online. doi: 10.1080/00401706.2021.1927849(link to open access pdf)

See Also

vcr.knn.newdata, classmap, silplot, stackedplot, e1071::svm

Examples

library(e1071)
set.seed(1); X <- matrix(rnorm(200 * 2), ncol = 2)
X[1:100, ] <- X[1:100, ] + 2
X[101:150, ] <- X[101:150, ] - 2
y <- as.factor(c(rep("blue", 150), rep("red", 50)))
cols <- c("deepskyblue3", "red")
plot(X, col = cols[as.numeric(y)], pch = 19)
# We now fit an SVM with radial basis kernel to the data:
set.seed(1) # to make the result of svm() reproducible.
svmfit <- svm(y~., data = data.frame(X = X, y = y),
scale = FALSE, kernel = "radial", cost = 10,
gamma = 1, probability = TRUE)
plot(svmfit$decision.values, col = cols[as.numeric(y)]); abline(h = 0)
# so the decision values separate the classes reasonably well.
plot(svmfit, data = data.frame(X = X, y = y), X.2~X.1, col = cols)
# The boundary is far from linear (but in feature space it is).
vcr.train <- vcr.svm.train(X, y, svfit = svmfit)
confmat.vcr(vcr.train)
stackedplot(vcr.train, classCols = cols)
classmap(vcr.train, "blue", classCols = cols)
classmap(vcr.train, "red", classCols = cols)

# For more examples, we refer to the vignette:
## Not run: 
vignette("Support_vector_machine_examples")

## End(Not run)