StatisticsML Knowledge
What’s your understanding of PCA and its inherent disadvantages?
Data ScientistMachine Learning Engineer
Microsoft
Meta
Akamai
Booking.com
Samsung
DocuSign
Answers
Anonymous
8 months ago
PCA, or principal component analysis, is a process to determine what are the most important features. I like to think about it in terms of eigenvalues. If you have a diagonal matrix, then the eigenvalues are the diagonal terms, and the eigenvectors are the basis vectors. In PCA, the principal components and their effects aren't as visible, but it is a process to determine them. If I recall correctly, PCA will return the absolute value of the eigenvalues, rather than the eigenvalue itself.
Interview question asked to Data Scientists and Machine Learning Engineers interviewing at LogMeIn, Motorola Solutions, Zendesk and others: What’s your understanding of PCA and its inherent disadvantages?.