-------- Original Message -------- Subject: Re: pGPA in same data differents results [R] Date: Wed, 22 Jul 2009 10:37:10 -0700 (PDT) From: Murat Maga <[email protected]> To: [email protected] References: <[email protected]> Dear Alejo, The coordinates in your first and third sample are very close to each other. Superimposition implementation in R is an iterative process, not an exact one. I presume those difference are within acceptable errors. The second set looks like scaled values of the 1st one. They are probably scaled by their respective centroid sizes. In fact if you plot those sets (1st and 2nd) independently I expect the relative position will be identical. What I am curious is why you get an unscaled values even though your SCALE parameter is set to True on the third set. Anyone? M morphmet wrote:
-------- Original Message -------- Subject: pGPA in same data differents results [R] Date: Tue, 21 Jul 2009 10:47:34 -0700 (PDT) From: Alejo . <[email protected]> To: [email protected] Dear Morphometricians, I want to start working with GM in R, and the first thing that I notice is that there are differences between superimposed coordinates (partial procrustes sup.) of same data using different approaches. I compared the registered coordinates of gorf.dat data using procGPA (shapes package, scale=F), and the functions in "Morphometrics with R", and the coordinates have very different values. Other thing is that procGPA report distances like "Kendall's Riemannian distance rho", and there is no documentation of how to convert to procrustes distance ( if they are different). Thanks a lot in advance! Here I paste a the coordinates of same indiv. but registered with prcoGPA or with functions of "Morphometric with R" of gorf.dat data procGPA(gorf.dat, scale=F) (scale=F Part. Proc. Sup.) [,1] [,2] [1,] -15.827220 115.57708 [2,] 14.952665 -107.48479 [3,] -35.791848 -76.45264 [4,] -33.230550 -43.55219 [5,] -29.636231 28.38585 [6,] -4.185892 97.61937 [7,] 44.839077 31.61518 [8,] 58.879998 -45.70786 pgpa(gorf.dat) (Function from book "Morphometrics wih R") [,1] [,2] [1,] -0.06524372 0.4917185 [2,] 0.06166891 -0.4572943 [3,] -0.15354677 -0.3244428 [4,] -0.14207133 -0.1845946 [5,] -0.12550966 0.1212242 [6,] -0.01606360 0.4151549 [7,] 0.19121899 0.1336318 [8,] 0.24954719 -0.1953976 procGPA(gorf.dat, scale=T) (scale=T, Full Proc. Sup) [,1] [,2] [1,] -15.995843 116.60296 [2,] 15.111564 -108.43879 [3,] -36.092444 -77.14230 [4,] -33.516190 -43.94813 [5,] -29.907041 28.63165 [6,] -4.246512 98.48805 [7,] 45.230989 31.90755 [8,] 59.415478 -46.10099
-- Replies will be sent to the list. For more information visit http://www.morphometrics.org
