DISCUSS ANYTHING & EVERYTHING: Money, Dating, Health, Internet, Legal, Jobs, People, Places and more

Hi,

I am quite new to this technique of Principal Component ****ysis but I wanted to know some solid fundamentals about it. So I started digging the web and got quite a large collection of study materials about it. Frankly speaking, I have got myself more confused than I was earlier on about it.

There are so many things involved in it and not just the simple procedure of finding the covariance matrix or the correlation matrix and their corresponding eigen ****ysis.

Well, to cut the story short, I would like to know if it is COMPLETELY NECESSARY to mean-center the data (I believe, it is also called normalization) if we are to do regression using the principal components obtained thereof.

Inputs, thoughts and clear ideas to this are warmly welcomed.

K