Say I have a situation where I do PCA with three components on a data set and the summed explained variance ratios of the three components is relatively low, say less than 0.5. If I were to then do k means clustering on these three components, would the resulting clusters be less reliable since the overall explained variances of the three components are so low?

  • $\begingroup$ Define "reliable". They can differ substantially from the result on the entire data, but that was the purpose of PCA in the first place. PCA distorts the data even when you'd use all components. $\endgroup$ – Anony-Mousse Aug 18 at 10:26

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.