midPCA
How does PCA behave with outliers?
Updated May 17, 2026
Short answer
PCA is sensitive to outliers because variance is heavily influenced by extreme values.
Deep explanation
Outliers distort covariance structure, shifting principal components toward extreme values. Robust scaling or robust PCA variants are used to mitigate this issue.
Real-world example
Fraud detection datasets with extreme transaction values.
Common mistakes
- Ignoring outlier treatment before PCA.
Follow-up questions
- How to handle outliers?
- Is PCA robust?