Data Classification with Decision Trees

Abstract

In this chapter, we will develop the theory for binary decision trees. Decision trees can be used to classify data, and fall into the Learning category in our Autonomous Learning taxonomy. Binary trees are easiest to implement because each node branches to two other nodes, or none. We will create functions for the Decision Trees and to generate sets of data to classify. Figure 7.1 shows a simple binary tree. Point “a” is in the upper left quadrant. The first binary test finds that its x value is greater than 1. The next test finds that its y value is greater than 1 and puts it in set 2. Although the boundaries show square regions, the binary tree really tests for regions that go to infinity in both x and y.