Home▸AI & Machine Learning▸Model Evaluation and Validation

📊 Model Evaluation and Validation

Fit a polynomial model to noisy data on a live 3D scatter plot, spin the k-fold cross-validation wheel, and watch the train/validation error bars trace out the classic overfitting curve.

AI & Machine Learning3DAdvanced60 FPS

A polynomial model fits a noisy 3D scatter plot in real time while a spinning k-fold cross-validation wheel and a live train/validation error chart reveal the bias-variance tradeoff as it happens.

🔬 What It Demonstrates

Raising model complexity always shrinks training error, but validation error follows a U-shape — improving, then rising again as the model starts memorizing noise instead of the underlying pattern.

🎮 How to Use

Slide model complexity to watch the fitted curve underfit, fit well, or overfit. Change the number of folds and the held-out fold to see k-fold cross-validation in action, or let it auto-cycle through every fold.

💡 Did You Know?

Because every point is validated exactly once across a full k-fold pass, cross-validation gives a far more reliable estimate of real-world performance than judging a model on a single lucky (or unlucky) train/test split.

⚙ Under the hood

Fit a polynomial model to noisy data on a live 3D scatter plot, spin the k-fold cross-validation wheel, and watch the train/validation error bars trace out the classic overfitting curve.

machine learningdata analysispolynomial regressioncross validationoverfittingmodel evaluationThree.js

3D · Three.js / WebGL renderer · 60 FPS target · runs fully client-side, no install