DSAverse

Initializing Sorting Algorithms...

Sorting

Trees

Graphs

Preparing interactive visualizations...

Back to Machine Learning

Decision Trees

Try every threshold on every feature, keep the question that leaves the purest groups, and repeat inside each group. Watch the scan, the boxes it carves, what an unlimited depth does, and the XOR case where choosing greedily goes wrong.

Builds onOverfitting and Underfitting Binary Search Tree

Every split searched in your browser · seeded two-moons and XOR data

Act 1 of 5 · Impuritystep 1 of 25

X[n, features] =[30, 2]

Speed

Thirty labelled points: 15 blue, 15 orange. A decision tree classifies by asking yes/no questions about one feature at a time. To pick a question it needs a score for how mixed a group is. Gini impurity is the chance that two points drawn at random (with replacement) disagree: 1 − Σ pₖ². Here 1 − (0.50² + 0.50²) = 0.50.

G = 1 - \left(\left(\tfrac{15}{30}\right)^2 + \left(\tfrac{15}{30}\right)^2\right) = 0.500

Decision Trees

Builds onOverfitting and Underfitting Binary Search Tree

Every split searched in your browser · seeded two-moons and XOR data

Act 1 of 5 · Impuritystep 1 of 25

X[n, features] =[30, 2]

Speed

G = 1 - \left(\left(\tfrac{15}{30}\right)^2 + \left(\tfrac{15}{30}\right)^2\right) = 0.500

DSAverse

Decision Trees

See also

Decision Trees

See also