Phase 3 · Machine learning · Models
Decision trees
Learn a flowchart of yes/no questions.
Open this lesson Create a free account
Why it matters
Decision trees learn rules you can read out loud: "if no order in 180 days and more than one support ticket, likely to churn." That makes them easy to explain to colleagues, and they handle mixed feature types, non-linear effects and interactions without any scaling. On their own they overfit easily; in groups (random forests, gradient boosting, next lesson) they're among the best models for business data, often beating neural networks on spreadsheets.
What you'll be able to do
- Read a decision tree as a flowchart of yes/no questions
- Measure how mixed a group is with Gini impurity
- Find the best split for a feature, the way tree algorithms do
- Control a tree's depth to keep it from memorizing the data
Before this lesson
Next in this phase: Random forests and boosting