Hi, I am doing a research on dimension reduction and one of the
steps involved requires me to project the L-dimensional data
orthogonally onto a d-space spanned by the d original coordinates
with the largest variances.
The L-dimensional data is actually a matrix with L cols. d-space
is a matrix which is smaller then the L-space. Which means that if
L = 15, d can be any value lesser than 15.
What I would like to know is how do I execute the statement
underlined above. Is there any maths formula I can go about
projecting a matrix orthogonally onto another matrix of a different
dimension?
Thanks.