References: Drawing the widest margin: support vector machines

Source material

Source material (conceptual spine):
• StatQuest with Josh Starmer: "Support Vector Machines (SVM)"
  Creator: Josh Starmer
  YouTube: https://www.youtube.com/watch?v=efR1C6CvhmE
• StatQuest with Josh Starmer: "Support Vector Machines Part 3: The Radial (RBF) Kernel"
  YouTube: https://www.youtube.com/watch?v=Qc5IyLW_hns
  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 “Support Vector Machines” anchors the maximum-margin idea, support vectors, and the soft margin, and the RBF kernel video anchors the kernel trick. StatQuest’s polynomial-kernel video (Part 2) covers the other common kernel.
• Microsoft’s ML-For-Beginners Classification module is the hands-on companion for fitting SVM classifiers in scikit-learn, including a reminder to scale features.

The squared-feature lifting example, the explicit feature-scaling gotcha contrasting with decision trees, and the framing of the kernel trick as a reusable “make it linear by lifting it” move are Clawdemy’s own.

Going deeper

• StatQuest with Josh Starmer. The main SVM video plus the polynomial-kernel and RBF-kernel parts. StatQuest builds the margin and kernel ideas with pictures, step by step.
• Microsoft ML-For-Beginners: Classification. Project-based lessons where you fit and tune an SVM in Python, scaling included.

Adjacent topics

• K-means clustering (the next lesson). The first unsupervised method: with no labels at all, group similar points together.
• The kernel trick beyond SVMs. The same lifting idea appears in kernel methods across machine learning. Recognizing “lift it into a space where it becomes linear” is worth more than the SVM alone.
• Bias and variance (Phase 4). The framework behind the soft-margin C parameter: wider margin trades a little training accuracy for better generalization.

Community discussion

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