morphmet
Mon, 14 Sep 2009 04:28:28 -0700
-------- Original Message -------- Subject: RE: PCA versus RWA Date: Sun, 13 Sep 2009 20:33:34 -0700 (PDT) From: F. James Rohlf <ro...@life.bio.sunysb.edu> Reply-To: ro...@life.bio.sunysb.edu Organization: Stony Brook University To: morphmet@morphometrics.org References: <4aad0bba.7020...@morphometrics.org> This paper gives some of the relationships. See equation (1) in particular.Rohlf, F. J. 1999. Shape statistics: Procrustes superimpositions and tangent spaces. Journal of Classification, 16:197-223.
------------------------ F. James Rohlf, Distinguished Professor Ecology & Evolution, Stony Brook University www: http://life.bio.sunysb.edu/ee/rohlf
-----Original Message----- From: morphmet [mailto:morphmet_modera...@morphometrics.org] Sent: Sunday, September 13, 2009 11:12 AM To: morphmet Subject: Re: PCA versus RWA -------- Original Message -------- Subject: Re: PCA versus RWA Date: Sat, 12 Sep 2009 18:18:32 -0700 (PDT) From: Andy Grass <adgr...@gmail.com> To: morphmet@morphometrics.org References: <4a9e5d21.8000...@morphometrics.org> How does one go about projecting the procrustes coordinates into the tangent space? This is actually a problem I've been wrestling with with my own data. I compute the procrustes and pw scores in IMP and then generally perform the PCA analysis in SAS and get two very different results between the straight procrustes PCA and the relative warps. On Wed, Sep 2, 2009 at 6:55 AM, morphmet <morphmet_modera...@morphometrics.org> wrote: > > > -------- Original Message -------- > Subject: RE: PCA versus RWA > Date: Tue, 1 Sep 2009 11:59:38 -0700 (PDT) > From: F. James Rohlf <ro...@life.bio.sunysb.edu> > Reply-To: <ro...@life.bio.sunysb.edu> > Organization: Stony Brook University > To: <morphmet@morphometrics.org> > References: <4a9d33ab.1090...@morphometrics.org> > > The partial warp scores represent a rigid rotation of the Procrustes > shape coordinates within the tangent space. Assuming one has taken the > step to first project the Procrustes shape coordinates into the > tangent space then the projections of the specimens onto the PCA axes > will be identical. > > That means there is no real advantage of computing partial warp scores > unless it is on interest to see how much of the shape variation is at > large versus small spatial scales. Well, perhaps one advantage - they > give you the proper number of shape variables so that one does not > have get the eigenvectors of an unnecessarily large matrix and then > discard the last 4 (or 7 for 3D data) eigenvectors that have > eigenvalues exactly equal to zero. > > ========================= > F. James Rohlf > Distinguished Professor, Stony Brook University > http://life.bio.sunysb.edu/ee/rohlf > > >> -----Original Message----- >> From: morphmet [mailto:morphmet_modera...@morphometrics.org] >> Sent: Tuesday, September 01, 2009 10:46 AM >> To: morphmet >> Subject: PCA versus RWA >> >> >> >> -------- Original Message -------- >> Subject: PCA versus RWA >> Date: Tue, 1 Sep 2009 06:20:00 -0700 (PDT) >> From: Alexandra Wegmann <inca...@yahoo.com> >> To: morphmet@morphometrics.org >> >> Hello >> >> Could anybody tell me when one should simply do a PCA (or CVA) on >> the >> procrusted shape coordinates and when one should do the PCA (or >> CVA) on >> the partial warp scores (relative warp analysis)? Your answer >> would be >> highly appreciated. >> >> >> >> Thanks in advance >> >> >> Alexandra >> Master student >> University of Zurich >> >> E-Mail: inca...@yahoo.com >> >> >> >> >> >> >> >> -- >> Replies will be sent to the list. >> For more information visit http://www.morphometrics.org > > > > > -- > Replies will be sent to the list. > For more information visit http://www.morphometrics.org > > -- Replies will be sent to the list. For more information visit http://www.morphometrics.org
-- Replies will be sent to the list. For more information visit http://www.morphometrics.org