Hands-On Recommendation Systems with Python

Decision trees

Decision trees are extremely fast and simple tree-based algorithms that branch out on features that result in the largest information gain. Decision trees, although not very accurate, are extremely interpretable.

We will not delve into the inner workings of the decision tree, but we will see it in action via a visualization:

Let's say we want to classify the Iris dataset using a decision tree. A decision tree performing the classification is shown in the preceding diagram. We start at the top and go deeper into the tree until we reach a leaf node.

For example, if the petal width of a flower is less than 0.8 cm, we reach a leaf ...