1
$\begingroup$

As title suggests, I've been wondering about how standardization works when trying to understand how Principal Component Analysis( PCA) works from this tutorial https://medium.com/analytics-vidhya/understanding-principle-component-analysis-pca-step-by-step-e7a4bb4031d9

Improve this question
$\endgroup$

1 Answer 1

Reset to default
0
$\begingroup$

YES

You can prove this by using properties of the mean and variance of a transformation, but the intuition is that subtracting the observed mean gives you a a new mean of zero (centers your data) and the dividing by the standard deviation compresses or expands the distribution to give a standard deviation of one.

Note that there is no requirement to standardize your data before you run PCA.

Improve this answer
$\endgroup$
2
  • $\begingroup$ Understood. Regarding the centering of data, how can I tell if I need to centralize it or not? Depending on how spread out the data is or are there other aspects to consider? $\endgroup$
    – Neri-kun
    Commented Feb 25, 2022 at 11:34
  • 1
    $\begingroup$ That’s a good topic to ask about in a new question. (Remember that Data Science, like most or all of the other Stacks, is strictly Q&A, not a discussion forum. It works best when each question has its own post, so that people with the same question in the future have a question where they can find an answer, rather than looking through discussions about related topics until they find the information they need.) $\endgroup$
    – Dave
    Commented Feb 25, 2022 at 11:40

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.