Decision trees to XGBoost and LightGBM — still the strongest baseline on tabular data, and interviewers know it.
10h · 8 lessons · 1 challenge
Before this: Regularization & Model Selection
If the data is tabular, gradient boosting is probably the answer, and being able to say why — and to tune it — is a real differentiator against candidates who reach for a neural network reflexively.
The module goes from a single tree through bagging and random forests to the boosting objective, then into practical tuning and interpretation with SHAP.
Grow a tree one split at a time against an opponent tree grown by the optimal criterion. See how far greedy gets you.
Worth 450 XP.
