MuhammadLab
LearningInteractiveBoosting

XGBoost — Gradient Boosted Trees (Explained)

Learn boosting intuition, why XGBoost works, and how regularization/shrinkage help.

What you’re seeing

Boosting builds an ensemble sequentially. Each step fits a small tree to the residuals (mistakes) of the current model, then adds a scaled correction using the learning rate η.

Top plot
Targets vs current predictions.
Bottom plot
Residuals + weak learner step.
Split line
Where the tiny tree divides x.

Prediction after m trees

Generate a simulation to populate this plot.

-0.080.210.500.791.08

Residuals and weak learner

Generate a simulation to populate this plot.

-0.080.210.500.791.08

Explanation

Generate a simulation to start.