Re: [R] Coefficient of Determination for nonlinear function
Dear Bert, dear Andy, thanks for your answers! I am quite aware that I do not fit a linear model, so r^2 in Pearson's sens is indeed meaningless. Instead, I am fitting an equation - or rather using an optimisation - were the experimentally derived point cloud (x1, x2, x3) should deliver something like 1 = f(x1, x2, x3). What I am trying to estimate is the quality of the fit. One thing I computed so far is the standard error of the equation (SEE) which is fine. My former question pointed in the direction of how I could compute a coefficient of determination to estimate a goodness of fit. Calling it r^2 may mislead but there must be something similar in nonlinear regressions. Thanks for your efforts, Uwe Am Freitag, den 04.03.2011, 11:44 -0500 schrieb Liaw, Andy: As far as I can tell, Uwe is not even fitting a model, but instead just solving a nonlinear equation, so I don't know why he wants a R^2. I don't see a statistical model here, so I don't know why one would want a statistical measure. Andy -Original Message- From: r-help-boun...@r-project.org [mailto:r-help-boun...@r-project.org] On Behalf Of Bert Gunter Sent: Friday, March 04, 2011 11:21 AM To: uwe.wolf...@uni-ulm.de; r-help@r-project.org Subject: Re: [R] Coefficient of Determination for nonlinear function The coefficient of determination, R^2, is a measure of how well your model fits versus a NULL model, which is that the data are constant. In nonlinear models, as opposed to linear models, such a null model rarely makes sense. Therefore the coefficient of determination is generally not meaningful in nonlinear modeling. Yet another way in which linear and nonlinear models fundamentally differ. -- Bert On Fri, Mar 4, 2011 at 5:40 AM, Uwe Wolfram uwe.wolf...@uni-ulm.de wrote: Dear Subscribers, I did fit an equation of the form 1 = f(x1,x2,x3) using a minimization scheme. Now I want to compute the coefficient of determination. Normally I would compute it as r_square = 1- sserr/sstot with sserr = sum_i (y_i - f_i) and sstot = sum_i (y_i - mean(y)) sserr is clear to me but how can I compute sstot when there is no such thing than differing y_i. These are all one. Thus mean(y)=1. Therefore, sstot is 0. Thank you very much for your efforts, Uwe -- Uwe Wolfram Dipl.-Ing. (Ph.D Student) __ Institute of Orthopaedic Research and Biomechanics Director and Chair: Prof. Dr. Anita Ignatius Center of Musculoskeletal Research Ulm University Hospital Ulm Helmholtzstr. 14 89081 Ulm, Germany Phone: +49 731 500-55301 Fax: +49 731 500-55302 http://www.biomechanics.de __ 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. -- Bert Gunter Genentech Nonclinical Biostatistics 467-7374 http://devo.gene.com/groups/devo/depts/ncb/home.shtml __ 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. Notice: This e-mail message, together with any attach...{{dropped:26}} __ 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.
[R] Coefficient of Determination for nonlinear function
Dear Subscribers, I did fit an equation of the form 1 = f(x1,x2,x3) using a minimization scheme. Now I want to compute the coefficient of determination. Normally I would compute it as r_square = 1- sserr/sstot with sserr = sum_i (y_i - f_i) and sstot = sum_i (y_i - mean(y)) sserr is clear to me but how can I compute sstot when there is no such thing than differing y_i. These are all one. Thus mean(y)=1. Therefore, sstot is 0. Thank you very much for your efforts, Uwe -- Uwe Wolfram Dipl.-Ing. (Ph.D Student) __ Institute of Orthopaedic Research and Biomechanics Director and Chair: Prof. Dr. Anita Ignatius Center of Musculoskeletal Research Ulm University Hospital Ulm Helmholtzstr. 14 89081 Ulm, Germany Phone: +49 731 500-55301 Fax: +49 731 500-55302 http://www.biomechanics.de __ 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.
[R] Coefficient of Determination for nonlinear function
Dear Subscribers, I did fit an equation of the form 1 = f(x1,x2,x3) using a minimization scheme. Now I want to compute the coefficient of determination. Normally I would compute it as r_square = 1- sserr/sstot with sserr = sum_i (y_i - f_i) and sstot = sum_i (y_i - mean(y)) sserr is clear to me but how can I compute sstot when there is no such thing than differing y_i. These are all one. Thus mean(y)=1. Therefore, sstot is 0. Thank you very much for your efforts, Uwe -- Uwe Wolfram Dipl.-Ing. (Ph.D Student) __ Institute of Orthopaedic Research and Biomechanics Director and Chair: Prof. Dr. Anita Ignatius Center of Musculoskeletal Research Ulm University Hospital Ulm Helmholtzstr. 14 89081 Ulm, Germany Phone: +49 731 500-55301 Fax: +49 731 500-55302 http://www.biomechanics.de __ 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.
Re: [R] How to plot Ellipsoid like function
Dear R-Users, I am trying to plot an ellipsoid like function that represents some physical threshold in its eigenvalue space. I am facing a few problems generating a figure I need for my thesis. A small example looks as follwos where the two contour3d plots do NOT overlay as desired so you may try plotting the surfaces one by one to see what I mean. # begin example require(rgl) require(misc3d) require(MASS); f - function(x, y, z){ chi0=-0.6603368 eps0=0.006590395 xi0=0.01117194 (x^2 + y^2 + z^2 - chi0*(x*y + x*z + y*z))/eps0^2 + (x + y + z)/xi0 } ff - function(x, y, z)x + y + z open3d() clear3d(all) bg3d(color=#88) light3d() x - seq(-.02,.02,len=20) # plot ellipsoid contour3d(f,1,x,x,x,color=#FF,alpha=0.5) # plot plane contour3d(ff,1,x,x,x,color=#FF,alpha=0.5) # plot data points spheres3d(c(-0.009379952, 0.007899338), c(-0.00879318, 0.00700924), c(-0.009009740, 0.007656409),radius=0.0005,color=#FF) # plot hydrostatic pressure line lines3d(c(-0.012, 0.012), c(-0.012, 0.012), c(-0.012, 0.012), col=#A8A8A8, lwd=4) # end example I have three questions regarding this problem and I hope you could help me. 1. How can I overlay the plan plotted using contours3d(ff, ...) and the ellipsoid plotted with contours3d(f, ...) 2. Instead of using spheres3d(...) I would love to use plot3d to obtain proper x, y and z coordinate axes. Is there a possibility to overly the contour3d() and line3d() commands with pot3d? Otherwise is there a possibility to plot proper coordinate axes with tics and such as usual R plots? 3. How can I save the scene to an image? pdf(...) ... dev.off() seems not to work on my machine. I am using Ubuntu on a 32 Bit Laptop. Thanks a million for your help! Uwe Am Montag, den 13.12.2010, 10:20 -0500 schrieb Duncan Murdoch: On 13/12/2010 10:13 AM, Uwe Wolfram wrote: I am currently trying to fit a tensorial function in its principal coorinate system. The function is given by: 1~(x1^2 + x2^2 + x3^2 - chi0*(x1*x2 + x1*x3 + x2*x3))/eps0^2 + (x1 + x2 + x3)/xi0 Where eps0 = 0.0066, chi0 = -0.66 and xi0 = 0.011 are obtained from experimental data using nls().I am able to plot the experimental points that delivered the parameters of the function. For my thesis, however, I need to overlay the fitted surface. So far I am using the following code which wonderfully plots the experimental points in 3D: === # from demo(bivar) require(rgl) require(misc3d) require(MASS); # New window open3d() # clear scene: clear3d(all) # setup env. That is, background, light and so on: bg3d(color=#88) light3d() # spheres at points in principal strain space #spheres3d(e1,e2,e3,radius=0.00025,color=#FF) # draws points alternatively plot3d(e1,e2,e3, col=#FF) === According to the examples on http://rgl.neoscientists.org/gallery.shtml I tried to overlay the point plot using surface3d. However, these were only functions of type y ~f(x1, x2). I think that the surface could be plotted if I could provide the gridpoints correctly. Using xyz.coords(1~(x1^2 + x2^2 + x3^2 - chi0*(x1*x2 + x1*x3 + x2*x3))/eps0^2 + (x1 + x2 + x3)/xi0, y = NULL, z = NULL) did unfortunately not solve the problem. Is there any function that can generate the surface for the given function such as ContourPlot3D in Mathematica. See ?misc3d::contour3d __ 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.
[R] How to plot Ellipsoid like function
Dear R-Users, I am currently trying to fit a tensorial function in its principal coorinate system. The function is given by: 1~(x1^2 + x2^2 + x3^2 - chi0*(x1*x2 + x1*x3 + x2*x3))/eps0^2 + (x1 + x2 + x3)/xi0 Where eps0 = 0.0066, chi0 = -0.66 and xi0 = 0.011 are obtained from experimental data using nls().I am able to plot the experimental points that delivered the parameters of the function. For my thesis, however, I need to overlay the fitted surface. So far I am using the following code which wonderfully plots the experimental points in 3D: === # from demo(bivar) require(rgl) require(misc3d) require(MASS); # New window open3d() # clear scene: clear3d(all) # setup env. That is, background, light and so on: bg3d(color=#88) light3d() # spheres at points in principal strain space #spheres3d(e1,e2,e3,radius=0.00025,color=#FF) # draws points alternatively plot3d(e1,e2,e3, col=#FF) === According to the examples on http://rgl.neoscientists.org/gallery.shtml I tried to overlay the point plot using surface3d. However, these were only functions of type y ~f(x1, x2). I think that the surface could be plotted if I could provide the gridpoints correctly. Using xyz.coords(1~(x1^2 + x2^2 + x3^2 - chi0*(x1*x2 + x1*x3 + x2*x3))/eps0^2 + (x1 + x2 + x3)/xi0, y = NULL, z = NULL) did unfortunately not solve the problem. Is there any function that can generate the surface for the given function such as ContourPlot3D in Mathematica. Thanks a million! Uwe -- Uwe Wolfram Dipl.-Ing. (Ph.D Student) __ Institute of Orthopaedic Research and Biomechanics Director and Chair: Prof. Dr. Anita Ignatius Center of Musculoskeletal Research Ulm University Hospital Ulm Helmholtzstr. 14 89081 Ulm, Germany Phone: +49 731 500-55301 Fax: +49 731 500-55302 http://www.biomechanics.de __ 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.
[R] Plotting 3d surfaces
Dear Subscribers, I am using R for quite a while nowadays on Ubuntu 10.04 LTS. I a using R for doing my statistics. Furthermore I am using it as a tool to generate the graphics for my publications. I am currently working on a project which involves nls-fits of three dimensional surfaces such as ellipsoids or even more complex. I have been searching R help and the manuals for a possibility to plot these surfaces. Is there any package in R that allows plotting something like this: http://uwwo.in-chemnitz.de/R-Problem/tsai-wu-principal-strain.pdf This plot was generated using Mathematicas ContourPlot3D function. I would love to be able to plot this in R. However, I could not find something which could this. Thanks a million for the help! Uwe -- Uwe Wolfram Dipl.-Ing. (Ph.D Student) __ Institute of Orthopaedic Research and Biomechanics Director and Chair: Prof. Dr. Anita Ignatius Center of Musculoskeletal Research Ulm University Hospital Ulm Helmholtzstr. 14 89081 Ulm, Germany Phone: +49 731 500-55301 Fax: +49 731 500-55302 http://www.biomechanics.de __ 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.