Decision Trees
Objectives
By the end, you'll be able to:
- Explain how a decision tree turns feature tests into predictions.
- Compare the main impurity criteria used to choose a split.
- Control model complexity with stopping rules and pruning.
Concepts Covered
- Tree anatomy: nodes, branches, and leaves
- Axis-aligned partitions of the feature space
- Classification error, Gini index, and entropy
- Split gain and leaf predictions
- Pre-pruning, post-pruning, and cost-complexity pruning
- Categorical variables and computational considerations