Online Robotics Classes for Kids in Brussels

1. What Is a Decision Tree?

A supervised algorithm that models decisions like humans: conditions first, then more specific conclusions. It supports:

• Classification — assign a category
• Regression — predict a continuous number
• Rule extraction — readable IF–THEN rules from training data

2. Structure

• Root node — first test (best starting feature)
• Internal nodes & branches — further questions and answer paths
• Leaf nodes — final prediction (class or value)

3. Splitting and Stopping

Splits use metrics such as Information Gain or Gini Impurity to make subsets purer. Growth stops when:

1. A node is pure (one class only)
2. Maximum depth is reached
3. Further splits add little purity

4. Example: Insurance Purchase

Root: Age > 30? If no → Buy = No. If yes → Income > $60,000? If yes → Buy = Yes; if no → Buy = No.

Paths become rules, e.g. IF Age > 30 AND Income > $60,000 THEN Buy = YES.

5. Explainable AI (XAI)

Unlike black-box models, every root-to-leaf path is an explicit rule. Auditors and stakeholders can see exactly why a prediction was made.