Support Vector Machines (SVM)

Support Vector Machines are a type of classification model that calculates a coordinate transformation of the data known as the kernel trick. On the transformed data, a hyperplane is used to divide the data for classification. In contrast to logistic regression, the shape of the classification boundary is not limited. To construct the boundary, a SVM selects a number of _"important" _points, the support vectors (hence the name).

SVMs can work with high-dimensonal datasets. They tend to be bit sensitive to parameters (type of kernel, regularization hyperparameter).

Code

library("e1071")

data(iris)

x <- subset(iris, select=-Species)

y <- iris[,"Species"]

model <- svm(x, y)

print(model)

summary(model)

# check accuracy

pred <- predict(model, x)

table(pred, y)

# color by class

# mark support vectors by +

plot(cmdscale(dist(iris[,-5])),

col = as.integer(iris[,5]),

pch = c("o","+")[1:150 %in% model$index + 1])