Re: [R] Proposal for New R List: Criticism? Comments?
On 09/10/04 03:54, Adaikalavan Ramasamy wrote: There is another issue to be considered. Currently you need to have the relevant packages installed before help.search() bring it up. My work around this is to install all available packages just in case the function I need is nestled in some non-standard packages. I also update them rather frequently. I do this too, at my search site (where frequently=monthly) and you can search functions only, and use Boolean search expressions and phrases. But right now the entire set of packages takes about 885 meg (if I'm reading du correctly), which is less than my very modest collection of digital photos, and a tiny fraction of a 3-year-old standard hard disk. In other words, it is no big deal to install all the packages if you have your own computer. Jon -- Jonathan Baron, Professor of Psychology, University of Pennsylvania Home page: http://www.sas.upenn.edu/~baron R search page: http://finzi.psych.upenn.edu/ __ [EMAIL PROTECTED] mailing list https://stat.ethz.ch/mailman/listinfo/r-help PLEASE do read the posting guide! http://www.R-project.org/posting-guide.html
Re: [R] Proposal for New R List: Criticism? Comments?
On Fri, 10 Sep 2004, Jonathan Baron wrote: On 09/10/04 03:54, Adaikalavan Ramasamy wrote: There is another issue to be considered. Currently you need to have the relevant packages installed before help.search() bring it up. My work around this is to install all available packages just in case the function I need is nestled in some non-standard packages. I also update them rather frequently. I do this too, at my search site (where frequently=monthly) and you can search functions only, and use Boolean search expressions and phrases. But right now the entire set of packages takes about 885 meg (if I'm reading du correctly), which is less than my very modest collection of digital photos, and a tiny fraction of a 3-year-old standard hard disk. In other words, it is no big deal to install all the packages if you have your own computer. I am seeing about 520Mb for all base + CRAN packages under 1.9.1, and it will be rather less under 2.0.0 as more parts are stored compressed. BioC is a lot larger. It is however, a BIG deal to install *all* the packages and am I currently 10 short since they depend on other software that I do not have a licence for or will not compile (and there are three others I cannot reinstall using current gcc). On AMD64 and Solaris there are several others, and something like 20 do not install on Windows. (I could use --install-fake as the CRAN checks do, but I have the almost complete set installed to test R changes, not test packages.) So I do see some merit in having a full-text search for R help available at some URL, as Jonathan has kindly provided. -- Brian D. Ripley, [EMAIL PROTECTED] Professor of Applied Statistics, http://www.stats.ox.ac.uk/~ripley/ University of Oxford, Tel: +44 1865 272861 (self) 1 South Parks Road, +44 1865 272866 (PA) Oxford OX1 3TG, UKFax: +44 1865 272595 __ [EMAIL PROTECTED] mailing list https://stat.ethz.ch/mailman/listinfo/r-help PLEASE do read the posting guide! http://www.R-project.org/posting-guide.html
Re: [R] ccf question
John Sibert wrote: Can someone please verify the interpretation of lag in the ccf function in the ts package. Suppose ccf(x,y). Do negative lags indicate that the events in x precede the events in y and positive lags indicate that events in y precede events in x? Thanks, John This you ncan answer for yourself by doing x - rnorm(200) ccf(x, lag(x), na.action=na.omit) and seing the value 1 at lag 1. Kjetil halvorsen John Sibert, Manager Pelagic Fisheries Research Program University of Hawaii at Manoa 1000 Pope Road, MSB 313 Honolulu, HI 96822 United States Phone: (808) 956-4109 Fax: (808) 956-4104 Washington DC Phone: (202) 861 2363 Fax: (202) 861 4767 PFRP Web Site: http://www.soest.hawaii.edu/PFRP/ email: [EMAIL PROTECTED] __ [EMAIL PROTECTED] mailing list https://stat.ethz.ch/mailman/listinfo/r-help PLEASE do read the posting guide! http://www.R-project.org/posting-guide.html -- Kjetil Halvorsen. Peace is the most effective weapon of mass construction. -- Mahdi Elmandjra __ [EMAIL PROTECTED] mailing list https://stat.ethz.ch/mailman/listinfo/r-help PLEASE do read the posting guide! http://www.R-project.org/posting-guide.html
RE: [R] loading Sjava
Apparently further explanation is in order, to correct some misimpressions. I was not aware that I was intruding on a private space. I was directed to the quoted URL for Sjava by the omegahat website, www.omegahat.org/RSJava, which states Currently, it is advisable to get the binary from Brian Ripley's Web site where the link is provided. After a refresh, I see that the page still says that. I had indeed read three readmes in the SJava distribution, as well as two different FAQs (one on the web and one in the distribution, and some of the documentation for usage. I did not read the readme in examples, putting that off, I think reasonably. I have not found the readme which you mention. The page www.omegathat.org/RSJava/FAQ.html states you will need version 1.2.0. My message mentioned R.dll because the FAQ discusses a problem if the installed version of R was not compiled as a shareable library. When I found R.dll I concluded that that problem is probably not applicable here. It's not true that I concluded that R.dll was the module referred to. If 1.6.2 is needed, as you suggest, I do not know yet if I will need to compile it with the flag --enable-R-shlib mentioned in the FAQ in reference to R.dll. If anyone already has a 1.6.2 build in binary, I would appreciate it. I do appreciate your help, Professor, and will pursue your suggestion vis a vis 1.6.2. Moderation of the pejoratives would be nice. -Original Message- From: Prof Brian Ripley [mailto:[EMAIL PROTECTED] Sent: Friday, September 10, 2004 9:04 AM To: Day, Roger Cc: [EMAIL PROTECTED] Subject: Re: [R] loading Sjava On Fri, 10 Sep 2004, Day, Roger wrote: I'm excited about SJava, and I'ld love to get it working, but can't get past loading the package. .First.lib fails on this statement: library.dynam(SJava, SJava, C:/PROGRA~1/R/rw1091/library) Error in dyn.load(x, as.logical(local), as.logical(now)) : unable to load shared library C:/PROGRA~1/R/rw1091/library/SJava/libs/SJava.dll: LoadLibrary failure: The specified module could not be found. Actually, SJava.dll is present there. But who said `SJava.dll' was the `specified module'? Hint: it is most likely not, but also you are trying to load a DLL built under R 1.6.2 into R 1.9.1 and that is likely to result in such a message. Advice Read the documentation, especially before posting as the posting guide asks. I'm using XP. Obtained the package from http://www.stats.ox.ac.uk/pub/bdr/SJava/ . Do learn to read ReadMe's! That one says SJava_0.65a.tar.gz modified sources SJava.zip build under R1.6.2 The sources here will build under R 1.6.x and R-devel (1.7.0-to-be). To install: Sjava.zip -- unzip in ...\rw1062\library. and note, not rw1091. That was a private area, and I have deleted it. There is a version of SJava in the public area http://www.stats.ox.ac.uk/pub/RWin, but do practice assiduously your new-found skills at reading ReadMe files. -- Brian D. Ripley, [EMAIL PROTECTED] Professor of Applied Statistics, http://www.stats.ox.ac.uk/~ripley/ University of Oxford, Tel: +44 1865 272861 (self) 1 South Parks Road, +44 1865 272866 (PA) Oxford OX1 3TG, UKFax: +44 1865 272595 __ [EMAIL PROTECTED] mailing list https://stat.ethz.ch/mailman/listinfo/r-help PLEASE do read the posting guide! http://www.R-project.org/posting-guide.html
[R] random seed
Hi, I have a R wrapper function that calls my C code via .C(). In my C code, I have been calling GetRNGstate() (and PutRNGstate()) once at the beginning (and the end) of the code. However, if I generate a random number within the R wrapper function (say via runif()), then my C code produces the exactly same numbers. If I don't generate a random number within the wrapper, it works fine. I wonder if there is something I need to do in order to prevent this problem. Any help would be appreciated. Best, Kosuke - Kosuke Imai Office: Corwin Hall 041 Assistant Professor Phone: 609-258-6601 Department of PoliticseFax: 973-556-1929 Princeton University Email: [EMAIL PROTECTED] Princeton, NJ 08544-1012 http://www.princeton.edu/~kimai __ [EMAIL PROTECTED] mailing list https://stat.ethz.ch/mailman/listinfo/r-help PLEASE do read the posting guide! http://www.R-project.org/posting-guide.html
RE: [R] Proposal for New R List: Criticism? Comments?
Dear Brian et al., Jonathan's search site is excellent -- I use it frequently -- and for some reason new users seem unaware of help.search(), which, despite the fact that it searches only in installed packages, I also find very useful. A couple of comments, however: First, if help pages from all packages were available at a central location -- e.g., at CRAN -- help.search() could have an option to search that location. Second, I still feel that it would be useful to provide some other way of searching the space of all available functions. One idea, which I mentioned in an earlier message on this thread, would be a keyword system (again, different from the current set of standard keywords). The keywords could be accessed by help.search() and also compiled into an index. Regards, John -Original Message- From: Prof Brian Ripley [mailto:[EMAIL PROTECTED] Sent: Friday, September 10, 2004 5:26 AM To: Jonathan Baron Cc: Adaikalavan Ramasamy; John Fox; R-help; 'Berton Gunter' Subject: Re: [R] Proposal for New R List: Criticism? Comments? On Fri, 10 Sep 2004, Jonathan Baron wrote: On 09/10/04 03:54, Adaikalavan Ramasamy wrote: There is another issue to be considered. Currently you need to have the relevant packages installed before help.search() bring it up. My work around this is to install all available packages just in case the function I need is nestled in some non-standard packages. I also update them rather frequently. I do this too, at my search site (where frequently=monthly) and you can search functions only, and use Boolean search expressions and phrases. But right now the entire set of packages takes about 885 meg (if I'm reading du correctly), which is less than my very modest collection of digital photos, and a tiny fraction of a 3-year-old standard hard disk. In other words, it is no big deal to install all the packages if you have your own computer. I am seeing about 520Mb for all base + CRAN packages under 1.9.1, and it will be rather less under 2.0.0 as more parts are stored compressed. BioC is a lot larger. It is however, a BIG deal to install *all* the packages and am I currently 10 short since they depend on other software that I do not have a licence for or will not compile (and there are three others I cannot reinstall using current gcc). On AMD64 and Solaris there are several others, and something like 20 do not install on Windows. (I could use --install-fake as the CRAN checks do, but I have the almost complete set installed to test R changes, not test packages.) So I do see some merit in having a full-text search for R help available at some URL, as Jonathan has kindly provided. -- Brian D. Ripley, [EMAIL PROTECTED] Professor of Applied Statistics, http://www.stats.ox.ac.uk/~ripley/ University of Oxford, Tel: +44 1865 272861 (self) 1 South Parks Road, +44 1865 272866 (PA) Oxford OX1 3TG, UKFax: +44 1865 272595 __ [EMAIL PROTECTED] mailing list https://stat.ethz.ch/mailman/listinfo/r-help PLEASE do read the posting guide! http://www.R-project.org/posting-guide.html
[R] Re: Bangdiwala
I am new to R and would be grateful if somebody could tell me how to access the Bangdiwala statistic after an agreement plot. Thanks, Martyn Dr. Martyn Sherriff Senior Lecturer, Dental Biomaterials Science, GKT Dental Institute, Floor 17, Guy's Tower, London Bridge, London SE1 9RT e-mail: [EMAIL PROTECTED] Tel. +44(0)-2071-881822 Fax. +44(0)-2071-881823 Departmental Home Page: http://tinyurl.com/2eotw Personal Home Page: http://tinyurl.com/kgkd Youth Rugby: www.fullerians.demon.co.uk The difference between winners and losers is that winners tell the jokes and the losers talk about the run of the ball. It's not over until the fat man whistles. __ [EMAIL PROTECTED] mailing list https://stat.ethz.ch/mailman/listinfo/r-help PLEASE do read the posting guide! http://www.R-project.org/posting-guide.html
Re: [R] Re: Bangdiwala
Martyn Sherriff wrote: I am new to R and would be grateful if somebody could tell me how to access the Bangdiwala statistic after an agreement plot. Thanks, Martyn Martyn, Since you didn't say so, I have to guess you are using package:vcd? If so, then ?agreementplot says: quote Value Invisibly returned, a list with components Bangdiwala the unweighted agreement strength statistic Bangdiwala.Weighted the weighted statistic weights the weigtht vector used. /quote So you should be able to do the following: library(vcd) data(SexualFun) ap - agreementplot(t(SexualFun)) ap$Bangdiwala (Please read the posting guide.) --sundar __ [EMAIL PROTECTED] mailing list https://stat.ethz.ch/mailman/listinfo/r-help PLEASE do read the posting guide! http://www.R-project.org/posting-guide.html
Re: [R] Proposal for New R List: Criticism? Comments?
Just finished updating and installing new packages from CRAN and BioConductor (including annotation data) and am happy to say that my R has just exceeded the 1 GB mark. On Fri, 2004-09-10 at 10:11, Jonathan Baron wrote: On 09/10/04 03:54, Adaikalavan Ramasamy wrote: There is another issue to be considered. Currently you need to have the relevant packages installed before help.search() bring it up. My work around this is to install all available packages just in case the function I need is nestled in some non-standard packages. I also update them rather frequently. I do this too, at my search site (where frequently=monthly) and you can search functions only, and use Boolean search expressions and phrases. But right now the entire set of packages takes about 885 meg (if I'm reading du correctly), which is less than my very modest collection of digital photos, and a tiny fraction of a 3-year-old standard hard disk. In other words, it is no big deal to install all the packages if you have your own computer. Jon __ [EMAIL PROTECTED] mailing list https://stat.ethz.ch/mailman/listinfo/r-help PLEASE do read the posting guide! http://www.R-project.org/posting-guide.html
[R] How to obtain a 95% envelope for the estimated cumulative risk function from bootstrap samples
Hi all, The hypothetical data is displayed as follows. ID time status 1 81 0 2 42 1 3 37 1 4 54 0 5 35 0 6 38 1 7 29 0 8 40 0 Question 1: How to obtain a 95% envelope for the estimated cumulative risk function from bootstrap samples? I guess the output of this step consists of the envelope and the estimated cumulative risk function. Question2: bootstrap process will be repeated for n times; then the average cumulative risk function was estimated as the median of the n empirical cumulative risk functions. My question is how to plot this so called Average Risk Function vs. time easily instead of using lines() and points() to draw the above step function? Thanks in advance. Rui __ [EMAIL PROTECTED] mailing list https://stat.ethz.ch/mailman/listinfo/r-help PLEASE do read the posting guide! http://www.R-project.org/posting-guide.html
[R] How to obtain a 95% envelope for the estimated cumulative risk function from bootstrap samples
Hi all, I am trying to replicate the results from a paper. The problems are in the setting of survival analysis. The hypothetical data is displayed as follows. ID time status 1 81 0 2 42 1 3 37 1 4 54 0 5 35 0 6 38 1 7 29 0 8 40 0 Question 1: How to obtain a 95% envelope for the estimated cumulative risk function from bootstrap samples? I guess the output of this step consists of the envelope and the estimated cumulative risk function. Question2: bootstrap process will be repeated for n times; then the average cumulative risk function was estimated as the median of the n empirical cumulative risk functions. My question is how to plot this so called Average Risk Function vs. time easily instead of using lines() and points() to draw the above step function? Thanks in advance. Rui __ [EMAIL PROTECTED] mailing list https://stat.ethz.ch/mailman/listinfo/r-help PLEASE do read the posting guide! http://www.R-project.org/posting-guide.html
[R] hclust, centroid
Does anyone know how hclust (stats) calculates centroid linkage if only a distance matrix can be used as the input? ...Tao __ [EMAIL PROTECTED] mailing list https://stat.ethz.ch/mailman/listinfo/r-help PLEASE do read the posting guide! http://www.R-project.org/posting-guide.html
Re: [R] Plotting irregular grid as image or persp
Thanks Deepanyan, On Sat, 28 Aug 2004, Deepayan Sarkar wrote: ... Yes, I think rgl would be the right tool for this. Even apart from the 3d acceleration issues, one of the problems with getting this in R would be that R doesn't do raster graphics, and I don't think hidden surface algorithms are very easy to implement in the R model. ... library(ncdf) # library(rgl) teapot-open.ncdf(teapot.nc) # wget http://www.maplepark.com/~drf5n/extras/teapot.nc edges-get.var.ncdf(teapot,tris) vertices-get.var.ncdf(teapot,locations) xy-vertices[c(1,2),] # this would be cooler with ?persp's trans3d plot(1:2,xlim=range(unlist(xy[1,])), ylim=range(xy[2,]),type='n') apply(edges,2,function(x){polygon(t(xy[,x]))}) I was playing around with this yesterday and got something similar (but more general). I didn't send it to you because I wasn't sure if that's what you wanted. Of course, I'm more familiar with the lattice version of trans3d, so it uses that. There are 2 versions, one using grid, one with base graphics. As I said, there are glitches due to faulty drawing order of the triangles. Shading is also possible (as implemented in wireframe), but those calculations are done in C code, so it would take a bit longer to carry over. library(grid) library(lattice) plotMesh.grid - function(l, z, rot.mat, dist = 0.1) ## rot.mat: 4x4 transformation matrix ## dist: controls perspective, 0 = none { x - ltransform3dto3d(l[,z], rot.mat, dist = dist) id - seq(length = ncol(x) / 3) ord - order(x[3, id * 3] + x[3, id * 3 - 1] + x[3, id * 3 - 2]) grid.newpage() xscale - range(x[1,]) yscale - range(x[2,]) md - max(diff(xscale), diff(yscale)) pushViewport(viewport(w = 0.9 * diff(xscale) / md, h = 0.9 * diff(yscale) / md, xscale = xscale, yscale = yscale)) id - as.vector(outer(1:3, (id[ord]-1) * 3, +)) grid.polygon(x = x[1,id], y = x[2,id], default.units = native, gp = gpar(fill = gray), id = rep(id[ord], each = 3)) } plotMesh.base - function(l, z, rot.mat, dist = 0.1, subset = TRUE) ## rot.mat: 4x4 transformation matrix ## dist: controls perspective, 0 = none { x - ltransform3dto3d(l[,z], rot.mat, dist = dist) id - seq(length = ncol(x) / 3) ord - order(x[3, id * 3] + x[3, id * 3 - 1] + x[3, id * 3 - 2]) xscale - range(x[1,]) yscale - range(x[2,]) plot(xscale, yscale, type = n) x - cbind(x, NA) id - as.vector(rbind(outer(1:3, (id[ord]-1) * 3, +), ncol(x))) polygon(x = x[1,id], y = x[2,id], col = gray) } rot.mat - ltransform3dMatrix(list(y = -30, x = 40)) plotMesh.grid(l, z, rot.mat, dist = 0) plotMesh.base(l, z, rot.mat, dist = 0) Those are nice -- I did want a varying color however, and needed to separate the calls to polygon: plotMesh.base-function(vertices,edges,col,rot.mat=diag(4),dist=0.1,...){ ## rot.mat a 4x4 homogeneous transformation matrix ## dist: controls perpective per lattice::ltransform3dto3d # rotate vertices-ltransform3dto3d(vertices,rot.mat,dist) xscale - range(vertices[1,]) yscale - range(vertices[2,]) plot(xscale, yscale, type = n) # find rough plot order ord-order(apply(edges,2,function(x){sum(vertices[3,x])})) if (length(col) == 1){ sapply(ord,function(x){ polygon(vertices[1,edges[,x]],vertices[2,edges[,x]],col=col,...)}) } else { sapply(ord,function(x){ polygon(vertices[1,edges[,x]],vertices[2,edges[,x]],col=col[x],...)}) } invisible(ord) } rot.mat - ltransform3dMatrix(list(z=45,y=30)) #;rot.mat plotMesh.base(l,z,rot.mat=rot.mat,col=rainbow(dim(z)[2]),lty=0) and the resultant image is: http://www.maplepark.com/~drf5n/images/teapot2.png I posted some really rough notes at http://www.maplepark.com/~drf5n/cgi-bin/wiki.cgi?RMeshVisualization Dave -- Dave Forrest [EMAIL PROTECTED](804)684-7900w [EMAIL PROTECTED] (804)642-0662h http://maplepark.com/~drf5n/ __ [EMAIL PROTECTED] mailing list https://stat.ethz.ch/mailman/listinfo/r-help PLEASE do read the posting guide! http://www.R-project.org/posting-guide.html