StackedML
Practice
Labs
Questions
Models
Pricing
Sign in
Questions
/
Unsupervised Learning
/
Dimensionality Reduction
/
When PCA fails
← Previous
Next →
588.
PCA Outlier Sensitivity
medium
PCA is sensitive to outliers. Why?
A
Outliers reduce the rank of the covariance matrix, causing some principal components to become undefined
B
PCA assigns outliers to their own components, inflating the number of components needed
C
Outliers have high leverage on the covariance matrix, pulling principal components toward them
D
Outliers violate the Gaussian assumption required for eigendecomposition to produce valid components
Sign in to verify your answer
← Back to Questions