Hands-On Machine Learning with C++

Decision tree algorithm overview

A decision tree is a supervised machine learning algorithm, based on how a human solves the task of forecasting or classification. Generally, this is a k-dimensional tree with decision rules at the nodes and a prediction of the objective function at the leaf nodes. The decision rule is a function that allows you to determine which of the child nodes should be used as a parent for the considered object. There can be different types of objects in the decision tree leaf—namely, the class label assigned to the object (in the classification tasks), the probability of the class (in the classification tasks), and the value of the objective function (in the regression task).

In practice, binary decision trees are ...