Hello! When using princomp() for principal components analysis, the resulting loadings matrix differs between R and the output from other programs (S-plus and SAS) The difference is that the sign of some columns in the loadings matrix. The absolute values, however, are the same. This sign difference also exists when comparing the princomp() results with the manually calculated eigen vectors using eigen(). Is this a bug in R? I am using R ver 1.5.1 Thanks in advance, Peter ==========================================Peter Vikstr?m Dept. of Economic History Ume? University SE-901 87 UME? E-mail: peter.vikstrom at ekhist.umu.se ========================================== -.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.- r-help mailing list -- Read http://www.ci.tuwien.ac.at/~hornik/R/R-FAQ.html Send "info", "help", or "[un]subscribe" (in the "body", not the subject !) To: r-help-request at stat.math.ethz.ch _._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._
On Thu, 3 Oct 2002, [iso-8859-1] Peter Vikström wrote:> Hello! > > When using princomp() for principal components analysis, the resulting > loadings matrix differs between R and the output from other programs (S-plus > and SAS) > The difference is that the sign of some columns in the loadings matrix. The > absolute values, however, are the same. > This sign difference also exists when comparing the princomp() results with > the manually calculated eigen vectors using eigen(). > Is this a bug in R?No. The results are only defined up to a sign change. -thomas -.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.- r-help mailing list -- Read http://www.ci.tuwien.ac.at/~hornik/R/R-FAQ.html Send "info", "help", or "[un]subscribe" (in the "body", not the subject !) To: r-help-request at stat.math.ethz.ch _._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._
On Thu, 3 Oct 2002 22:19:48 +0200, an incredible array of electrons randomly cascading around the Universe collided and turned into words spewed forth by Peter Vikstr?m: Peter> Hello! Peter> Peter> When using princomp() for principal components analysis, the resulting Peter> loadings matrix differs between R and the output from other programs (S-plus Peter> and SAS) Peter> The difference is that the sign of some columns in the loadings matrix. The Peter> absolute values, however, are the same. Peter> This sign difference also exists when comparing the princomp() results with Peter> the manually calculated eigen vectors using eigen(). Peter> Is this a bug in R? Peter> I am using R ver 1.5.1 Signs of principal components are arbitrary. It isn't uncommon at all to have two programs give the same absolute value pc's but differ in sign. -.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.- r-help mailing list -- Read http://www.ci.tuwien.ac.at/~hornik/R/R-FAQ.html Send "info", "help", or "[un]subscribe" (in the "body", not the subject !) To: r-help-request at stat.math.ethz.ch _._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._