Vs (PCA & LDA)
Description
| Feature | PCA (Principal Component Analysis) | LDA (Linear Discriminant Analysis) |
|---|---|---|
| Primary Goal | Reduces dimensions by finding axes (PCs) that capture the most variation in the overall dataset. | Reduces dimensions by finding axes (LDs) that maximize the separability between known categories. |
| Method Type | Unsupervised (doesn't use category labels). | Supervised (requires known category labels). |
| Axis Ranking | Axes are ranked by the amount of total variance explained: • PC1 accounts for the most variation. • PC2 accounts for the second most. | Axes are ranked by their ability to separate categories: • LD1 accounts for the most variation between categories. • LD2 accounts for the second most. |