Model Selection Tournament Game

Tournament Dashboard

Total Score0

Current Round0 / 7

Models Remaining0

Best Composite—

Decision Time00:00

Streak0

1. Tournament Setup

Choose the challenge

Dataset scenario

Difficulty

Game mode

Identify the learning task

Deployment scenario

2. Select Tournament Algorithms

0 selected

Select at least four compatible algorithms. Expert mode rewards diversity and penalises unsuitable entrants.

3. Build the Evaluation Pipeline

Preprocessing, validation, metrics, and weights

Data preparation

Missing values

Numerical scaling

Categorical encoding

Outlier handling

Feature selection

Class balancing

Validation design

Test split 20%

Validation split 20%

Cross-validation 5

Noise stress 10%

Random seed 42

Primary metrics

Multi-criteria selection weights

Performance35

Speed15

Memory10

Interpretability15

Stability15

Fairness10

Weights are normalised automatically. Deployment scenarios can modify their influence.

4. Hyperparameter Arena

Shared tournament tuning

Tree and ensemble settings

Trees / estimators180

Maximum depth10

Minimum split8

Optimisation settings

Learning rate0.08

Epochs100

Batch size64

Model-specific settings

Neighbours7

Regularisation0.35

Dropout0.20

5. Tournament Arena

Tournament coach

Configure a dataset, choose the correct task, select algorithms, and start the tournament.

6. Live Model Comparison

Rank	Algorithm	Validation	Test	CV Mean	CV Variation	Train Time	Prediction	Memory	Interpretability	Fairness	Generalisation Gap	Composite	Status
No tournament results yet.

7. Select the Champion

Final decision and justification

Your chosen champion

Decision confidence

Explain your choice

🏆 Tournament result

Champion

Final score

Game Options and Accessibility

Stored in this browser

Guided explanations

Sound effects

Reduced animation

High-contrast indicators

Auto-advance rounds

Keyboard shortcuts

Keyboard controls

Press S to start, N for the next round, P to pause, and R to reset.

Selection lesson

The strongest score is not always the best deployment choice. Balance performance, risk, and constraints.

Simulation note

Results are educational simulations. They demonstrate trade-offs rather than training production models.

Progress is automatically preserved after completed rounds.