Labs
Interactive, build-it-in-your-browser explainers for ideas that are easier to see than to read about. Everything runs live on a fixed seed — no pre-rendered pictures, just real algorithms and sliders.
How Decision Trees Carve Up Space
Grow a tree one split at a time, watch leaves turn into probabilities, see why a single tree is high-variance, and how averaging many into a random forest produces a smooth boundary.
Finding the 1%: Trees & Forests on Imbalanced Data
Why a 99:1 class split makes models confidently ignore the rare class, and how threshold tuning, class weighting, RUS/ROS bagging, and SMOTE rebuild a useful decision boundary.
Trees That Fix Their Own Mistakes
How gradient boosting adds trees one at a time to correct residual errors, why stochastic row and column sampling adds variety, and why a boosted ensemble can perfectly interpolate a time series yet completely fail to extrapolate it.