Tutorial
Jun 26, 20266 min read
0
Unsupervised Learning and the Curse of Dimensionality
Supervised learning trains on (X, y) pairs. Unsupervised learning gets only X — no labels, no correct answers. The algorithm must find structure on its own. Thi…
~/blog/tutorials/machine-learning
Unsupervised
Tutorial
Jun 26, 20269 min read
0
Feature Selection vs Feature Extraction
Dimensionality reduction takes two fundamentally different paths. Feature selection keeps a subset of original features — you end up with columns you can name a…
Tutorial
Jun 26, 20269 min read
0
PCA: Geometric and Math Intuition
Principal Component Analysis finds a new coordinate system for your data — one aligned with the directions of maximum spread. The axes (principal components) ar…
Tutorial
Jun 26, 20268 min read
0
PCA: Eigendecomposition
Post 03 showed that PCA finds the direction u that maximizes $\mathbf{u}^T C \mathbf{u}$, and that a diagonal direction outperforms the original axes. This post…