Tree Depth Balancer Game

Model Laboratory

Intermediate

Balanced Fit Mission

Find a simple tree with strong validation accuracy and a small generalisation gap.

Challenge setup

Game mode

Dataset

Difficulty

Tree controls

Maximum depth 4

Minimum samples per split 4

Minimum samples per leaf 2

Maximum leaf nodes 12

Pruning strength 20%

Split criterion

Feature limit

Data laboratory

Training sample percentage 70%

Label noise 10%

Random seed

Animation

Experience options

Sound effects

Dark chart theme

High contrast

Guided explanations

Live Model Scorecard

Not trained

Training accuracy

—

Fit on known samples

Validation accuracy

—

Primary balance signal

Test accuracy

—

Unseen performance

Generalisation gap

—

Train minus validation

Complexity score

—

Depth, nodes and pruning

F1 score

—

Precision-recall balance

Precision

—

Positive prediction quality

Recall

—

Positive coverage

Tree nodes

—

— leaves

CV mean

—

Five-fold score

Prediction speed

—

Estimated samples/sec

Resource cost

—

Relative model budget

Interactive Model Visualiser

Animated decision tree

Synthetic decision boundary

Class A
Class B
Misclassified

Configure your first tree

Choose a depth, train the model, and inspect how simplicity changes fit and generalisation.

Plotly Analysis Studio

Attempt History

#	Depth	Validation	Gap	Diagnosis	Score
Train a tree to record an attempt.

Achievements

0 / 8

Results and Accessibility

Keyboard: T train, H hint, U undo, R reset