Re: Help on Partial Least Squares Aalgorithm

From: Beorne (mdtorre_at_freemail.it)
Date: 02/16/05 

Date: 16 Feb 2005 07:00:28 -0800

To say the truth, I'm witing an application structured in two parts:
1) dimensionality reduction (or feature extraction) of the problem (I
use this for visualization too)
2) regression (linear, quadratic, radial basis function network, feed
forward network)

I'd like to split PLS1 in these two parts, in particular I'd like to
find the subspace W (surely a function of X and y) of PLS algorithm
"directly". I think that If I know this function I'll understand better
the pls too (For me is a bit hard to understand pls from the iterative
PLS2 algorithms).
It is possible that in this way my computation is more complex and
more lengthy, but I need it to better fit in the application and to
better understand PLS in my "mental framework".

Quantcast