midPCA
What is the difference between PCA and feature selection?
Updated May 17, 2026
Short answer
PCA creates new features; feature selection keeps original features.
Deep explanation
Feature selection chooses subset of original variables, while PCA transforms them into new orthogonal components. PCA is unsupervised and focuses on variance, not interpretability.
Real-world example
Selecting important genes vs compressing gene expression data.
Common mistakes
- Using PCA when interpretability is required.
Follow-up questions
- Which preserves interpretability?
- Which is better for compression?