Hi all, I found that the PCA gave chaotic results when there are big changes in a few data points. Are there "improved" versions of PCA in R that can help with this problem? Please give me some pointers... Thank you! [[alternative HTML version deleted]]
Michael: On Thu, Apr 19, 2012 at 9:20 PM, Michael <comtech.usa at gmail.com> wrote:> Hi all, > > I found that the PCA gave chaotic results when there are big changes in a > few data points.Yup.> > Are there "improved" versions of PCA in R that can help with this problem?Yup. Consult the "Robust" task view on CRAN. You probably also ought to consult a local statistician, as understanding R/R covariance matrix estimates requires more statistical training than you've had.. But if not, the task view is the right R resource to start with. -- Bert> > Please give me some pointers... > > Thank you! > > ? ? ? ?[[alternative HTML version deleted]] > > ______________________________________________ > R-help at r-project.org mailing list > https://stat.ethz.ch/mailman/listinfo/r-help > PLEASE do read the posting guide http://www.R-project.org/posting-guide.html > and provide commented, minimal, self-contained, reproducible code.-- Bert Gunter Genentech Nonclinical Biostatistics Internal Contact Info: Phone: 467-7374 Website: http://pharmadevelopment.roche.com/index/pdb/pdb-functional-groups/pdb-biostatistics/pdb-ncb-home.htm
You can also have a look at the pcaMethods package on Bioconductor. Kevin On Thu, Apr 19, 2012 at 11:20 PM, Michael <comtech.usa@gmail.com> wrote:> Hi all, > > I found that the PCA gave chaotic results when there are big changes in a > few data points. > > Are there "improved" versions of PCA in R that can help with this problem? > > Please give me some pointers... > > Thank you! > > [[alternative HTML version deleted]] > > ______________________________________________ > R-help@r-project.org mailing list > https://stat.ethz.ch/mailman/listinfo/r-help > PLEASE do read the posting guide > http://www.R-project.org/posting-guide.html > and provide commented, minimal, self-contained, reproducible code. >-- Kevin Wright [[alternative HTML version deleted]]
I could not reply directly to the initial thread with the same title. There are two sorts of Robust PCA, those that were devised before the recent string of Low Rank approaches and then the new set of algorithms that provide robust PCA in light of sparse but potentially large errors/outliers (typically the sort of outliers that break normal PCA). These recent algorithms initially come from some of the folks involved in compressive sensing. I am keeping a list of all these new solvers here in the Matrix Factorization Jungle Page @ https://sites.google.com/site/igorcarron2/matrixfactorizations Most are written in Matlab and should not need be too difficult to translate into R. To get a sense of what these new Robust PCA techniques can do, a friend and I apply on different YouTube videos, you can see some of the entries listed here: http://nuit-blanche.blogspot.com/p/its-cai-cable-and-igors-adventures-in.html ------------------------ Igor Carron, Ph.D. http://nuit-blanche.blogspot.com [[alternative HTML version deleted]]