These are just four of the most popular classifiers out there, but there are many more to choose from. Although some classification mechanisms perform better on some types of datasets than others, it can be hard to develop an intuition for exactly the ones they are suitable for. In order to help with this, we will be examining the efficacy of our four classifiers on four different two-dimensional made-up datasets—each with a vastly different optimal decision boundary. In doing so, we will learn more about the characteristics of each classifier and have a better sense of the kinds of data they might be better suited for.
The four datasets are depicted in Figure 9.11: