define N as the number of eigenvectors needed to capture at least "99.9" of the variation in d

How find find?

[Arifuzzaman Joy]

N=5;

Answer 1

[John D'Errico]

Lets see. Do you know how to compute what is the total variation? I hope so! Call it totalVar.

What is 99.9% of that number? totalVar*0.999.

Now look at the eigenvalues. Start summing them up from the largest down, (you might even use cumsum) until you just exceed totalVar*0.999.

I don't see the problem.

[Walter Roberson]

eigenvalues can be negative though ? Or complex ?

Answer 2

[the cyclist]

In general, you should provide more context and information about your question, so that we do not have to guess at what you are doing.

However, I am going to make a guess here. If you have the Statistics and Machine Learning Toolbox, then I think you can use the pca function. It has output called explained, which is the fraction of variation that each principal component captures.

You might also want to see my answer to this question, which gives a lot of explanation about how to use and interpret pca results.