Mastering Predictive Analytics with R

Chapter 5. Support Vector Machines

In this chapter, we are going to take a fresh look at nonlinear predictive models by introducing support vector machines. Support vector machines, often abbreviated as SVMs, are very commonly used for classification problems, although there are certainly ways to perform function approximation and regression tasks with them. In this chapter, we will focus on the more typical case of their role in classification. To do this, we'll first present the notion of maximal margin classification, which presents an alternative formulation of how to choose between many possible classification boundaries and differs from approaches such as maximum likelihood, which we have seen thus far. We'll introduce the related idea of ...