Mathematical formula for PCA

[Varun K Mohan]

Hello,

Is there a simple mathematical formula to define PCA? If so, please let me know.

Thanks,

--
Varun Mohan

[Kourosh Meshgi]

Hi
I guess the Wikipedia page is more than complete: http://en.wikipedia.org/wiki/Principal_component_analysis
But if you like to hear it simple, it is just a 3 step process (for n dimensional data matrix):
1- Perform the singulat value decomposition on your data, this will give you eigenvectors v and eigenvalues lambda
2- Sort eigenvectors regarding their corresponding eigenvalue in descending fashion
3- Choose the m top vectors (m<n), those will be your principle component vectors (the vectors can best describe your data in their direction)

--
You received this message because you are subscribed to the Google Groups "Face Recognition Research Community" group.
To post to this group, send email to face...@googlegroups.com.
To unsubscribe from this group, send email to face-rec+u...@googlegroups.com.
For more options, visit this group at http://groups.google.com/group/face-rec?hl=en.