Home▸Machine Learning & Neural Networks▸Random Forest Visualization - Interactive Ensemble Learning

🧪 Random Forest Visualization - Interactive Ensemble Learning

Watch a real random forest train in 3D: bootstrap-sampled decision trees split a scatter of points by Gini impurity, vote by majority, and paint a live decision surface — with feature subsampling and out-of-bag accuracy you can toggle live.

Machine Learning & Neural Networks2DModerate60 FPS📱 Mobile-adapted

This simulation grows a real random forest in 3D instead of animating a canned diagram of one. A ring of literal trees each trains on its own bootstrap-resampled draw of the data, splitting nodes only on a random subset of the two available features and choosing thresholds that minimize Gini impurity — the same recursive CART procedure scikit-learn runs under the hood. The forest floor at the center renders the ensemble's actual majority-vote decision surface, recomputed every time you retrain, alongside a live out-of-bag accuracy estimate computed from the points each tree never saw during its own training.

⚙ Under the hood

This simulation illustrates the concept of random forests and ensemble learning. It visually represents how multiple decision trees combine to make predictions, improving accuracy.

Random ForestEnsemble Learning

2D · HTML5 Canvas 2D · 60 FPS target · runs fully client-side, no install