Decision Trees

Learn how decision trees classify observations by recursively splitting data.

Decision trees classify data by asking a series of simple yes-or-no questions. Each branch splits the data until the model reaches a prediction.

Goal

Train a decision tree on the Iris dataset and explore how max_depth affects the model.

How Decision Trees Work

A decision tree starts at the root and splits data based on feature values. For example, it might ask whether petal length is greater than 2.5 cm, then continue splitting until each leaf represents a class label.

Why max_depth Matters

max_depth limits how many splits the tree can make. A shallow tree may underfit, while a very deep tree may memorize the training data. In the notebook below, try different values and compare accuracy.

Practice

Try It Yourself

Open the practice lab to complete the starter code in the notebook.

Knowledge Check

Quick Quiz

What does max_depth control in a decision tree?

Summary

Key Takeaways

  • Decision trees split data using feature-based rules.
  • Tree depth controls model complexity.
  • Deep trees can overfit training data.