Supervised Machine Learning with Python

An intuitive example – decision tree

A decision tree will start out with all of the data, iteratively making splits until each leaf has maximized its purity or some other stopping criteria is met. In this example, we will start out with three samples. The tree learns that splitting on the color feature will be our most informative step towards maximizing its leaf purity. So, that's the first thing to note. The first split is the most informative split that will best segment the data into two pieces. As shown in the following diagram, the potato class is isolated on the left by splitting on color. We have perfectly classified the potato. However, the other two samples still need to be split. So, the tree learns that, if it's orange and round, ...