Bias-Variance Balancing Game

Game controls

Round 1

Game mode

Difficulty

Algorithm

Model complexity5

Polynomial degree 5

Regularisation0.08

Training epochs120

Extra irrelevant features1

Dataset laboratory

Pattern

Samples

Random seed

Noise level0.18

Outliers3

Train %

Valid %

Test %

Model arena

Drag the complexity slider, fit the model, and inspect generalisation.

Balanced candidate 02:00

● Train ◆ Validation ■ Test — Prediction ·· True pattern

Training MSE

—

Fit to known examples

Validation MSE

—

Model selection signal

Test MSE

Hidden

Revealed after submission

Generalisation gap

—

Validation minus training

Cross-validation

—

Mean ± spread

Bias estimate

—

Variance estimate

—

Overfitting risk

—

Parameters

—

Effective model capacity

Score

Best: 0

Current objective

Trials: 0 / 12

Find the validation minimum. Keep the generalisation gap small while avoiding excessive complexity.

Generate a dataset, then adjust model complexity. Your feedback will appear here.

Bias–variance dashboard

Bias pressureLow

Variance pressureLow

Generalisation riskLow

Fit a model to receive a diagnosis.

Learning curves and trade-offs

Saved model comparison

Model	Algorithm	Complexity	Train	Validation	Bias	Variance	Gap	CV
No saved candidates yet.

Achievements

Bias Buster Variance Tamer Perfect Balance Simplicity Champion Cross-Validation Master Noise Resistant Regularisation Specialist Early-Stopping Expert

Personal best

Stored locally

Current run

Total score

Rounds cleared

Completed challenges

Display, accessibility, and export

Show true pattern

Show test points

Show residual lines

Show uncertainty

Sound effects

High contrast

Reduced motion

←→ adjust complexity F fit Enter submit H hint N new dataset