You should also have a look at the pcaMethods package (on
Bioconductor). I did some code optimization for the NIPALS algorithm
in that package which sped up the algorithm by at least a factor of
two.
Kevin Wright
On Wed, Apr 30, 2008 at 10:56 PM, steven wilson <[EMAIL PROTECTED]> wrote:
> Dear
> I've been trying to find a way to improve (if
> possible) the efficiency of the code, especially when the number of
> components to calculate is higher than 100.
> pca.nipals <- function(X, ncomp, iter = 50, toler =
sqrt(.Machine$double.eps))
> # X...data matrix, ncomp...number of components,
>
Some comments
1) 'Writing R Extensions' has a chapter about looking into bottlenecks
('profiling'). Had you given a working example, I might have shown you
some results.
One thing which is clearly suboptimal is to grow components (in your case
by cbind) rather than allocate them initially an
Dear list members;
The code given below corresponds to the PCA-NIPALS (principal
component analysis) algorithm adapted from the nipals function in the
package chemometrics. The reason for using NIPALS instead of SVD is
the ability of this algorithm to handle missing values, but that's a
different
4 matches
Mail list logo
