References: Turning weak learners strong: boosting

Source material

Source material (conceptual spine):
• StatQuest with Josh Starmer: "AdaBoost"
  Creator: Josh Starmer
  YouTube: https://www.youtube.com/watch?v=LsK-xG1cLYA
• StatQuest with Josh Starmer: "Gradient Boost Part 1: Regression Main Ideas"
  YouTube: https://www.youtube.com/watch?v=3CC4N4z3GJc
  Channel / site: https://statquest.org/
  License: as published on StatQuest's public YouTube channel (link-out only)

Source material (hands-on companion):
• Microsoft: "ML For Beginners" (Classification module)
  Repository: https://github.com/microsoft/ML-For-Beginners
  License: MIT

Clawdemy provides original notes, summaries, and quizzes derived from this material
for educational purposes. All rights to the original videos and curriculum remain
with their creators.

What this lesson draws from each source

• StatQuest’s “AdaBoost” anchors the re-weighting flavor (focus on misclassified examples, weight each learner’s vote), and “Gradient Boost Part 1” anchors the residual-chasing flavor. StatQuest has multi-part series on both for readers who want the full mechanics.
• Microsoft’s ML-For-Beginners Classification module is the hands-on companion for fitting boosted models in scikit-learn.

The explicit bagging-versus-boosting contrast table, the residual-shrinking worked trace, and the callback to gradient descent as the same engine are Clawdemy’s own connective tissue.

Going deeper

• StatQuest with Josh Starmer. The AdaBoost video and the four-part Gradient Boost series (regression main ideas, the math, classification). StatQuest also has a dedicated XGBoost series.
• XGBoost documentation. The most widely used gradient-boosting library. Its “Introduction to Boosted Trees” page is a clear, practical companion once the intuition here is solid.
• Microsoft ML-For-Beginners: Classification. Project-based lessons where you fit boosted classifiers in Python.

Adjacent topics

• Support vector machines (the next lesson). A return to drawing a single boundary, with the principle of maximizing the margin between classes.
• Bias and variance (Phase 4). The framework that makes “boosting cuts bias, bagging cuts variance” precise, and the basis for choosing between them on a given problem.
• Early stopping. The standard guard against boosting’s overfitting: stop adding trees when validation error stops improving. A practical companion to the learning-rate tuning named here.

Community discussion

None selected for this lesson. Boosting is well covered by the StatQuest, XGBoost, and Microsoft resources above. If a canonical discussion surfaces, it will be added at the next review.