Max Margin Classification Using SVMs

SVM is an algorithm for supervised learning that solves both classification and regression problems. However, SVM is most commonly used in classification problems, so, for the purposes of this chapter, we will focus on SVM as a binary classifier. The goal of SVM is to determine the best location of a hyperplane that create a class boundary between data points plotted on a multidimensional space. To help clarify this concept, refer to Figure 3.20.

Figure 3.20: Hyperplane (blue) separating the circles from the squares in three dimensions

In Figure 3.20, the squares and circles are observations in the same DataFrame that represent different classes. In this figure, the hyperplane is depicted by a semi-transparent blue boundary lying between the circles and squares that separate the observations into two distinct classes. In this example, the observations are said to be linearly separable.

The location of the hyperplane is determined by finding...