[R] simulation
Hello I would like to simulate datasets in the following way: x - rpois(999, 2000) y - sum(exp(rgamma(x, scale=2, shape=0.5))) The problem is, that by calling y I just get 1 value back and not 999 values. Can anyone help me? Thanks! Brigitte [[alternative HTML version deleted]] __ R-help@stat.math.ethz.ch 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] bivariate normal distribution in likelihood
On Tue, 1 May 2007, Robert A LaBudde wrote: At 11:32 PM 4/30/2007, Deepankar wrote: Hi all, I am trying to do a maximum likelihood estimation. In my likelihood function, I have to evaluate a double integral of the bivariate normal density over a subset of the points of the plane. For instance, if I denote by y the y co-ordinate and by x, the x co-ordinate then the area over which I have to integrate is the following set of points, A: A = {x=0 y 3x+10} I have used the following code to calculate this double integral: x - rmvnorm(10, mean=me, sigma=sig) c - nrow(x) x1 - x[ ,1] x2 - x[ ,2] e1 - as.numeric(x2 3*x1 + 10 x10) p1 - sum(e1)/c In this code, I provide the mean and covariance while drawing the bivariate random normal variables and get p1 as the required answer. The problem is that I have to draw at least 100,000 bivariate random normals to get a reasonable answer; even then it is not very accurate. Is there some other way to do the same thing more accurately and more efficiently? For instance, can this be done using the bivariate normal distribution function pmvnorm? Also feel free to point our errors if you see one. Simple random sampling is a poor way to evaluate an integral (expectation). It converges on the order of 1/sqrt(N). Which is no worse than other general schemes in high dimensions or without smoothness assumptions on the integrand. Stratified random sampling would be better, as it converges on the order of 1/N. Your reference for this result, please. (As stated it is untrue, so I presume the stratification scheme depends on N and there are smoothness assumptions.) (BTW, the reference for the results I quote is 'Stochastic Simulation' (1987).) We have not been told 'me' and 'sig', and depending on their values it is quite possible that importance sampling would do a great deal better than sampling from the specified bivariate normal. Even better is product Gauss-Hermite quadrature which will give a very accurate answer with a few dozen points. This is a correlated bivariate normal, and product quadrature methods can be arbitrarily bad for such integrals (as people find out for mixed linear models). What you can do is transform to a pair of uncorrelated normals, and for a set of the form A as given this transforms to a similar form. And for that an iterated integral can be done easily as the inner integral over y will just be a call to pnorm. Specifically, there is another normal z such that x and z are independent and y = z + a*x for some a. Then A = {x 0 z (3-x)*x + 10} can be integrated over z conditional on x and then over x. Robert A. LaBudde, PhD, PAS, Dpl. ACAFS e-mail: [EMAIL PROTECTED] Least Cost Formulations, Ltd.URL: http://lcfltd.com/ 824 Timberlake Drive Tel: 757-467-0954 Virginia Beach, VA 23464-3239Fax: 757-467-2947 Vere scire est per causas scire __ R-help@stat.math.ethz.ch 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. -- 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 __ R-help@stat.math.ethz.ch 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] Perpendicular symbol in plotmath?
That's fantastic, guys! Thank you very much. Paul's solution will definitely suffice until the perpendicular symbol is implemented in plotmath. -Matt On 1 May 2007, at 00:40, Paul Murrell wrote: Hi Matthew Neilson wrote: Thanks for your response, Gabor. That works quite nicely. The documentation states that it is not possible to mix and match Hershey fonts with plotmath symbols. My *ideal* scenario would be to write the perpendicular symbol as a subscript (specifically, I would like to have \epsilon_{\perp} as an axis label). I have searched the help archive, and it turned up the following post from 2002: http://tinyurl.com/2m8n9c which explains a way of faking subscripts when using the Hershey fonts, though it does have several drawbacks. Have things moved on in the last five years, or is this still the best known solution? Unfortunately, you still cannot use Hershey fonts with plotmath (just lacking implementation). Also, the perpendicular symbol is not implemented in plotmath (yet). In this case though, there may be a possible workaround. Try the following ... plot(1, ann=FALSE) title(ylab=expression(epsilon[\136]), family=symbol) The plain text character \136 gets drawn using the symbol font and the perpendicular symbol is character 94 (Octal 136) in the Adobe Symbol Encoding and in the Windows symbol font encoding so this works for PDF, on Windows, and on X11 (though I had to switch to a single-byte encoding to get my system to pick up the symbol font). The drawback with this solution is that anything that is NOT a special mathematical symbol in the expression will come out in Greek letters. Paul Many thanks for your help, -Matt On Sat Apr 28 17:35 , 'Gabor Grothendieck' [EMAIL PROTECTED] sent: Its available in the Hershey fonts: plot(0, 0, type = n) text(0, 0, A \\pp B, vfont = c(serif, plain)) On 4/28/07, Matthew Neilson [EMAIL PROTECTED] wrote: Hey, Does anyone know of an equivalent to the LaTeX \perp (perpendicular) symbol for adding to R plots? Parallel is easy enough (||), but I haven't been able to find a way of adding perpendicular. The plotmath documentation doesn't mention how to do it, so I'm inclined to think that it doesn't exist - but surely there must be some way of achieving the desired result, right? Any help will be much appreciated, -Matt __ R-help@stat.math.ethz.ch 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-help@stat.math.ethz.ch 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. -- Dr Paul Murrell Department of Statistics The University of Auckland Private Bag 92019 Auckland New Zealand 64 9 3737599 x85392 [EMAIL PROTECTED] http://www.stat.auckland.ac.nz/~paul/ -- Matthew Neilson University of Strathclyde Department of Mathematics Livingstone Tower 26 Richmond Street Glasgow G1 1XH Tel : + 44(0)141 548 4559 e-mail : [EMAIL PROTECTED] __ R-help@stat.math.ethz.ch 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] Odp: simulation
Hi [EMAIL PROTECTED] napsal dne 01.05.2007 09:03:46: Hello I would like to simulate datasets in the following way: x - rpois(999, 2000) y - sum(exp(rgamma(x, scale=2, shape=0.5))) You computed sum of your 999 values. Regardless of how many values are summed the result is always only one number. Did not you want cumsum? Regards Petr The problem is, that by calling y I just get 1 value back and not 999 values. Can anyone help me? Thanks! Brigitte [[alternative HTML version deleted]] __ R-help@stat.math.ethz.ch 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-help@stat.math.ethz.ch 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] simulation
maybe you're looking for something like this: x - rpois(999, 2000) y - numeric(length(x)) for (i in seq_along(x)) y[i] - sum(exp(rgamma(x[i], scale = 2, shape = 0.5))) I hope it helps. Best, Dimitris Dimitris Rizopoulos Ph.D. Student Biostatistical Centre School of Public Health Catholic University of Leuven Address: Kapucijnenvoer 35, Leuven, Belgium Tel: +32/(0)16/336899 Fax: +32/(0)16/337015 Web: http://med.kuleuven.be/biostat/ http://www.student.kuleuven.be/~m0390867/dimitris.htm Quoting Thür Brigitte [EMAIL PROTECTED]: Hello I would like to simulate datasets in the following way: x - rpois(999, 2000) y - sum(exp(rgamma(x, scale=2, shape=0.5))) The problem is, that by calling y I just get 1 value back and not 999 values. Can anyone help me? Thanks! Brigitte [[alternative HTML version deleted]] __ R-help@stat.math.ethz.ch 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. Disclaimer: http://www.kuleuven.be/cwis/email_disclaimer.htm __ R-help@stat.math.ethz.ch 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] thousand separator (was RE: weight)
Liaw, Andy wrote: Looks very neat, Gabor! I just cannot fathom why anyone who want to write numerics with those separators in a flat file. That's usually not for human consumption, and computers don't need those separators! Andy The world is stranger than you think... (One of my favourite gripes is the used of localized decimal separators in .csv files, which used to be a perfectly well-defined data format, but is now nonportable across locales. Or even between programs on the same machine: AFAIR, SAS reads and writes the US format in all locales, and SPSS uses the local format...) __ R-help@stat.math.ethz.ch 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] bivariate normal distribution in likelihood
Robert A LaBudde wrote: At 11:32 PM 4/30/2007, Deepankar wrote: Hi all, I am trying to do a maximum likelihood estimation. In my likelihood function, I have to evaluate a double integral of the bivariate normal density over a subset of the points of the plane. For instance, if I denote by y the y co-ordinate and by x, the x co-ordinate then the area over which I have to integrate is the following set of points, A: A = {x=0 y 3x+10} I have used the following code to calculate this double integral: x - rmvnorm(10, mean=me, sigma=sig) c - nrow(x) x1 - x[ ,1] x2 - x[ ,2] e1 - as.numeric(x2 3*x1 + 10 x10) p1 - sum(e1)/c In this code, I provide the mean and covariance while drawing the bivariate random normal variables and get p1 as the required answer. The problem is that I have to draw at least 100,000 bivariate random normals to get a reasonable answer; even then it is not very accurate. Is there some other way to do the same thing more accurately and more efficiently? For instance, can this be done using the bivariate normal distribution function pmvnorm? Also feel free to point our errors if you see one. Simple random sampling is a poor way to evaluate an integral (expectation). It converges on the order of 1/sqrt(N). Stratified random sampling would be better, as it converges on the order of 1/N. Even better is product Gauss-Hermite quadrature which will give a very accurate answer with a few dozen points. Or, use the mvtnorm package. It has pmvnorm, which does the integrals over rectangular regions. You'll need a pretransformation to use it for the problem at hand, though (from (x,y) to (x,y-3x)). __ R-help@stat.math.ethz.ch 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] simulation
Dimitris Rizopoulos wrote: maybe you're looking for something like this: x - rpois(999, 2000) y - numeric(length(x)) for (i in seq_along(x)) y[i] - sum(exp(rgamma(x[i], scale = 2, shape = 0.5))) Or use sapply, sapply(x, function(x) sum(exp(rgamma(x[i], scale = 2, shape = 0.5)) ) or even replicate(999, sum(exp(rgamma(rpois(1,2000), scale = 2, shape = 0.5)) ) __ R-help@stat.math.ethz.ch 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] simulation
Thats exactly what I am looking for! Thanks for your help! -Ursprüngliche Nachricht- Von: Peter Dalgaard [mailto:[EMAIL PROTECTED] Gesendet: Dienstag, 1. Mai 2007 11:46 An: Dimitris Rizopoulos Cc: Thür Brigitte; r-help@stat.math.ethz.ch Betreff: Re: [R] simulation Dimitris Rizopoulos wrote: maybe you're looking for something like this: x - rpois(999, 2000) y - numeric(length(x)) for (i in seq_along(x)) y[i] - sum(exp(rgamma(x[i], scale = 2, shape = 0.5))) Or use sapply, sapply(x, function(x) sum(exp(rgamma(x[i], scale = 2, shape = 0.5)) ) or even replicate(999, sum(exp(rgamma(rpois(1,2000), scale = 2, shape = 0.5)) ) *** [EMAIL PROTECTED] scanned this email for malicious content and found it to be clean *** __ R-help@stat.math.ethz.ch 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] impossible to open SPSS file
David Barron gave me a good tip that I had not seen: Have a look at FAQ 2.16 in the R for Windows FAQ. Knut Krueger wrote: My workaround for that problem is to build the data sheet in Excel and import it to SPSS and R. Or to export the SPSS sheet to CSV (or maybe if it is possible to Excel) and import it to R SPSS does not externalize the data format :-( Regards Knut We don't have SPSS so unfortunately this work around is not possible. Thank you both, Frank -- .. Dr. Frank Thomas FTR Internet Research 93110 Rosny-sous-Bois France __ R-help@stat.math.ethz.ch 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] Problem with the installation of install R on Sun Solaris
This file contains any messages produced by compilers while running configure, to aid debugging if configure makes a mistake. It was created by R configure 2.4.1, which was generated by GNU Autoconf 2.59. Invocation command line was $ ./configure ## - ## ## Platform. ## ## - ## hostname = ardsass1d uname -m = i86pc uname -r = 5.10 uname -s = SunOS uname -v = Generic_118855-36 /usr/bin/uname -p = i386 /bin/uname -X = System = SunOS Node = ardsass1d Release = 5.10 KernelID = Generic_118855-36 Machine = i86pc BusType = unknown Serial = unknown Users = unknown OEM# = 0 Origin# = 1 NumCPU = 3 /bin/arch = i86pc /usr/bin/arch -k = i86pc /usr/convex/getsysinfo = unknown hostinfo = unknown /bin/machine = unknown /usr/bin/oslevel = unknown /bin/universe = unknown PATH: /usr/sbin PATH: /usr/bin PATH: /opt/sfw/bin PATH: /usr/openwin/bin ## --- ## ## Core tests. ## ## --- ## configure:1805: checking build system type configure:1823: result: i386-pc-solaris2.10 configure:1831: checking host system type configure:1845: result: i386-pc-solaris2.10 configure:2397: checking for pwd configure:2415: found /usr/bin/pwd configure:2428: result: /usr/bin/pwd configure:2435: checking whether builddir is srcdir configure:2443: result: yes configure:2448: checking for working aclocal configure:2452: result: found configure:2461: checking for working autoconf configure:2465: result: found configure:2474: checking for working automake configure:2478: result: found configure:2487: checking for working autoheader configure:2491: result: found configure:2500: checking for working makeinfo configure:2508: result: missing configure:2517: checking for gawk configure:2533: found /opt/sfw/bin/gawk configure:2543: result: gawk configure:2553: checking for egrep configure:2563: result: egrep configure:2568: checking whether ln -s works configure:2572: result: yes configure:2620: checking for ranlib configure:2647: result: : configure:2663: checking for bison configure:2692: result: no configure:2663: checking for byacc configure:2692: result: no configure:2704: checking for ar configure:2733: result: no configure:2756: checking for a BSD-compatible install configure:2811: result: /opt/sfw/bin/install -c configure:2846: checking for sed configure:2865: found /usr/xpg4/bin/sed configure:2877: result: /usr/xpg4/bin/sed configure:2896: checking for less configure:2914: found /usr/bin/less configure:2926: result: /usr/bin/less configure:2948: checking for perl configure:2966: found /usr/bin/perl configure:2978: result: /usr/bin/perl configure:2989: checking whether perl version is at least 5.004 configure:3000: result: yes configure:3073: checking for dvips configure:3106: result: no configure:3118: checking for tex configure:3151: result: no configure:3163: checking for latex configure:3196: result: no configure:3210: WARNING: you cannot build DVI versions of the R manuals configure:3217: checking for makeindex configure:3250: result: no configure:3262: checking for pdftex configure:3295: result: no configure:3307: checking for pdflatex configure:3340: result: no configure:3354: WARNING: you cannot build PDF versions of the R manuals configure:3361: checking for $(SHELL) configure:3394: result: no configure:3361: checking for $(top_srcdir)/tools/missing configure:3394: result: no configure:3361: checking for makeinfo configure:3394: result: no configure:3361: checking for makeinfo configure:3394: result: no configure:3462: checking for unzip configure:3480: found /usr/bin/unzip configure:3492: result: /usr/bin/unzip configure:3507: checking for zip configure:3525: found /usr/bin/zip configure:3537: result: /usr/bin/zip configure:3552: checking for gzip configure:3570: found /usr/bin/gzip configure:3582: result: /usr/bin/gzip configure:3599: checking for firefox configure:3632: result: no configure:3599: checking for mozilla configure:3632: result: no configure:3599: checking for netscape configure:3632: result: no configure:3599: checking for galeon configure:3632: result: no configure:3599: checking for kfmclient configure:3632: result: no configure:3599: checking for opera configure:3632: result: no configure:3599: checking for gnome-moz-remote configure:3632: result: no configure:3599: checking for open configure:3632: result: no configure:3642: WARNING: I could not determine a browser configure:3655: checking for acroread configure:3688: result: no configure:3655: checking for acroread4 configure:3688: result: no configure:3655: checking for xpdf configure:3673: found /opt/sfw/bin/xpdf configure:3685: result: /opt/sfw/bin/xpdf configure:3749: checking for gcc configure:3778: result: no configure:3829: checking for cc configure:3858: result: no configure:3871: checking for cc configure:3917: result: no configure:3970: checking for cl configure:3999: result: no configure:4013: error: no acceptable C compiler found in $PATH See `config.log'
Re: [R] Concepts question: environment, frame, search path
On 01/05/2007 12:29 AM, Graham Wideman wrote: Folks: I'd appreciate if someone could straighten me out on a few concepts which are described a bit ambiguously in the docs. 1. data.frame: Refan p84: 'A data frame is a list of variables of the same length with unique row names, given class data.frame.' I probably don't need to point out how opaque that is! Which manual are you looking at? The reference index (refman.pdf)? It doesn't usually include statements like that; they are usually found in the Introduction to R (R-intro.pdf) or the R Language Definition (R-lang.pdf). But since the refman is just a collection of man pages, it might be in there somewhere. And since the manuals do get updated, that statement may not be present in the current release. (I did a quick search of the source, and couldn't spot it, but my search might have failed because of line breaks, strange formatting, or looking in the wrong place.) By the way, it's generally best to cite the section name where you found a quote, because the pagination varies from system to system. Even better would be to give a URL to the online HTML version at http://cran.r-project.org/manuals.html. For future reference, if you are suggesting a change, it's best to cite the line number in the source at https://svn.r-project.org/R/trunk/doc/manual in the *.texi files or https://svn.r-project.org/R/trunk/src/library/*/man/*.Rd for man pages, and send such suggestions to the R-devel list. Anyhow, key question: Some places in the docs seem pretty firm that a data.frame is basically a 2-D array with: a) named rows and b) columns whose items within a column be of uniform data type. Elsewhere, it seems like a data.frame can be a collection of arbitrary variables. The former interpretation is correct. Since the variables all have the same length, things like df[i, j] make sense: they choose the i'th entry from the j'th variable (according to the refan definition), or the i'th row, j'th column (according to the 2-D array interpretation. 2. environment --- Refman p122: Environments consist of a frame, or collection of named objects, and a pointer to an enclosing environment. Is the or here explaining parenthetically that a frame is a collection of named objects, or is separating too alternative structures for an environment? The former. If the former, does this imply that a frame can contain arbitrary variables? Yes, but a frame isn't an R object, it's a concept that appears in descriptions, e.g. part of an environment, or the local variables created during function evaluation, etc. And pointer? Is that a type of thing in R? No, there are no pointers in R. There are a couple of tricks to fake them (e.g. environment objects aren't copied when assigned, you just get a new reference to the same environment; this allows you to construct something like a pointer by wrapping an object in an environment), but I don't recommend using these routinely. 3. R search path; attach() The R search path appears to hold the list of collections of data (my term) that can be accessed by a users' commands. Refman p27 tells that search path can hold items that are data.frame, list, environment or R data file (on disk). Yet R-intro p28 describes attach() as taking a directory name argument. What is the concept directory in this context? I haven't read the preceding pages carefully, but that looks like an error. The usual argument to attach is a package name, and what gets attached is an environment holding the exports from the package. Packages are stored in directories in the file system, so maybe that's what the author of that line had in mind. Duncan Murdoch __ R-help@stat.math.ethz.ch 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] adding column to a matrix
l have the following dataset and would like to calculate the actual survival time by if censoring time survival time then actual survival time =survival time else its= censoring time. treatmentgrp strata censoringTimesurvivalTime censoring actualsurvivaltim [1,] 1 1 1.012159 1137.80922 0 [2,] 2 2 32.971439 247.21786 0 [3,] 2 1 85.758253 797.04949 0 [4,] 1 1 16.99917178.92309 0 l used matrix to genarate the data thanks in advance - [[alternative HTML version deleted]] __ R-help@stat.math.ethz.ch 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] intersect filled.contour and polygon
On Mon, 30 Apr 2007, Pedro Mardones wrote: Dear R users; Is there any way to intersect a filled contour image and a polygon? My problem is that I want to create a kriging map and the boundaries of my map are given by the coordinates of the polygon. So far I can superompose the polygon in the filled.contour image but I don't know how to get rid of the contour image outside of the polygon boundaries. Two possibilities seem to be present: define your polygon as a hole inside the bounding box of the image, with a link between the polygon and the bounding box, and fill it; or, perhaps better, only make kriging predictions for the polygon. Since you haven't given a code example, it isn't obvious how you are making the kriging predictions, but the choice of the newdata locations ought to decide - and they do not need to be a full grid, or even a regular grid at all. If you need to follow this up, please consider using the R-sig-geo list. Any hint will be appreciated PM __ R-help@stat.math.ethz.ch 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. -- Roger Bivand Economic Geography Section, Department of Economics, Norwegian School of Economics and Business Administration, Helleveien 30, N-5045 Bergen, Norway. voice: +47 55 95 93 55; fax +47 55 95 95 43 e-mail: [EMAIL PROTECTED] __ R-help@stat.math.ethz.ch 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] to draw a smooth arc
Paulo Barata wrote: Dear R-list members, I would like to draw a smooth arc. I can draw an arc parametrically, but this produces an arc too coarse, even allowing for different increments in sequence t in the example below. Function symbols (graphics) does produce a smooth circle, but it cannot produce an arc. Please see the following example, drawing complete circles: plot(-5:5,-5:5,type='n') ## draws circle with function symbols (package graphics) ## - inner circle is very smooth: symbols(0,0,circles=2,add=TRUE) ## draws circle parametrically - outer circle is too coarse: pi - 4*atan(1) t - seq(0,2*pi,0.02) lines(4*cos(t),4*sin(t)) Package plotrix has a function draw.arc, but arcs produced with this function are also either too coarse or too polygonal, depending on the number of polygons used to approximate the arc. Is there a way to harness the characteristics of function symbols (graphics) to draw a smooth arc, not just a complete circle? Hi Paulo, I may be misunderstanding you, but have you tried to increase the number of segments in the arc using the n argument? draw.arc(1,1,1,n=100) Jim __ R-help@stat.math.ethz.ch 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] Odp: adding column to a matrix
Hi see ?ifelse ifelse(censoringsurvival, survival, censoring) Regards Petr [EMAIL PROTECTED] [EMAIL PROTECTED] napsal dne 01.05.2007 13:25:18: l have the following dataset and would like to calculate the actual survival time by if censoring time survival time then actual survival time =survival time else its= censoring time. treatmentgrp strata censoringTimesurvivalTime censoring actualsurvivaltim [1,] 1 1 1.012159 1137.80922 0 [2,] 2 2 32.971439 247.217860 [3,] 2 1 85.758253 797.049490 [4,] 1 1 16.99917178.92309 0 l used matrix to genarate the data thanks in advance - [[alternative HTML version deleted]] __ R-help@stat.math.ethz.ch 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-help@stat.math.ethz.ch 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] to draw a smooth arc
On Tue, 1 May 2007, Jim Lemon wrote: Paulo Barata wrote: Dear R-list members, I would like to draw a smooth arc. I can draw an arc parametrically, but this produces an arc too coarse, even allowing for different increments in sequence t in the example below. Function symbols (graphics) does produce a smooth circle, but it cannot produce an arc. Please see the following example, drawing complete circles: plot(-5:5,-5:5,type='n') ## draws circle with function symbols (package graphics) ## - inner circle is very smooth: symbols(0,0,circles=2,add=TRUE) ## draws circle parametrically - outer circle is too coarse: pi - 4*atan(1) t - seq(0,2*pi,0.02) lines(4*cos(t),4*sin(t)) Package plotrix has a function draw.arc, but arcs produced with this function are also either too coarse or too polygonal, depending on the number of polygons used to approximate the arc. Is there a way to harness the characteristics of function symbols (graphics) to draw a smooth arc, not just a complete circle? Hi Paulo, I may be misunderstanding you, but have you tried to increase the number of segments in the arc using the n argument? draw.arc(1,1,1,n=100) Put it another way, drawing arcs is not a primitive in the R graphics system but drawing circles is. So there is no low-level way to draw an arc of a circle except via line segments. (Quite a few graphics devices draw circles via line segments, but not all and vector-graphics systems like postscript can often do better.) -- 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 __ R-help@stat.math.ethz.ch 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] Independent contrasts from lme with interactions
Ken, Take a look at the just released contrast package. Max -Original Message- From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED] On Behalf Of Ken Nussear Sent: Monday, April 30, 2007 6:12 PM To: r-help@stat.math.ethz.ch Subject: [R] Independent contrasts from lme with interactions Hi All, I've been searching the help archives but haven't found a workable solution to this problem. I'm running an lme model with the following call: lme.fnl - lme(Max ~ S + Tr + Yr + Tr:Yr, random = ~1 |TID) anova(lme.fnl) numDF denDF F-value p-value (Intercept) 1 168 19255.389 .0001 S 1 168 5.912 0.0161 Tr 2 11615.919 .0001 Yr 1 16877.837 .0001 Tr:Yr 2 16847.584 .0001 summary(lme.fnl) Linear mixed-effects model fit by REML Data: NULL AIC BIClogLik 580.6991 613.5399 -281.3496 Random effects: Formula: ~1 | TID (Intercept) Residual StdDev: 0.3697006 0.5316062 Fixed effects: Max ~ S + Tr + Yr + Tr:Yr Value Std.Error DF t-value p-value (Intercept) -13.5681 113.2623 168 -0.119793 0.9048 SM 0.21870.0957 168 2.284605 0.0236 TrT97 1375.5897 164.0060 116 8.387434 0. TrT98 2890.9462 455.3497 116 6.348848 0. Yr 0.00990.0567 168 0.174005 0.8621 TrT97:Yr -0.68830.0821 168 -8.384798 0. TrT98:Yr -1.44630.2279 168 -6.347310 0. Correlation: (Intr) SM TrT97 TrT98 Yr TT97:Y SM0.067 TrT97-0.691 -0.049 TrT98-0.248 -0.001 0.171 Yr -1.000 -0.067 0.691 0.248 TrT97:Yr 0.691 0.048 -1.000 -0.171 -0.691 TrT98:Yr 0.248 0.001 -0.171 -1.000 -0.248 0.171 Standardized Within-Group Residuals: Min Q1 Med Q3 Max -2.19017911 -0.58108001 -0.04983642 0.57323031 2.39811353 Number of Observations: 291 Number of Groups: 119 I'm specifically interested in differences of in the differences between my treatment groups (3) and Year (Yr), and importantly in the interaction. Normally I'm used to running independent contrast analysis to explore these differences, but I'm not sure how to extract this information using lme. Can anyone point me in the right direction? Thanks Ken [[alternative HTML version deleted]] __ R-help@stat.math.ethz.ch 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. -- LEGAL NOTICE\ Unless expressly stated otherwise, this messag...{{dropped}} __ R-help@stat.math.ethz.ch 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] Problem with the installation of install R on Sun Solaris
R has a file INSTALL which asks you to read R-admin.html if you have a problem. That manual explains that you need a C (preferably C99) and a Fortran compiler. This extensive posting merely says that you don't have a C compiler in your path. That's not something we can help you with. (Lots of other tools are missing as well.) On Tue, 1 May 2007, Vipin Singhal wrote: This file contains any messages produced by compilers while running configure, to aid debugging if configure makes a mistake. It was created by R configure 2.4.1, which was generated by GNU Autoconf 2.59. Invocation command line was $ ./configure [...] configure:4013: error: no acceptable C compiler found in $PATH -- 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 __ R-help@stat.math.ethz.ch 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] thousand separator (was RE: weight)
--- Liaw, Andy [EMAIL PROTECTED] wrote: Looks very neat, Gabor! I just cannot fathom why anyone who want to write numerics with those separators in a flat file. That's usually not for human consumption, and computers don't need those separators! Andy It' often a case of taking what you can get. I've seem myself taking formatted numbers from report intended for reading and then cutting and pasting them into a text editor. From: Gabor Grothendieck That could be accomplished using a custom class like this: library(methods) setClass(num.with.junk) setAs(character, num.with.junk, function(from) as.numeric(gsub(,, , from))) ### test ### Input - A B 1,000 1 2,000 2 3,000 3 DF - read.table(textConnection(Input), header = TRUE, colClasses = c(num.with.junk, numeric)) str(DF) On 4/30/07, Liaw, Andy [EMAIL PROTECTED] wrote: Still, though, it would be nice to have the data read in correctly in the first place, instead of having to do this kind of post-processing afterwards... Andy From: Bert Gunter Nothing! My mistake! gsub -- not sub -- is what you want to get 'em all. -- Bert Bert Gunter Genentech Nonclinical Statistics -Original Message- From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED] On Behalf Of Marc Schwartz Sent: Monday, April 30, 2007 10:18 AM To: Bert Gunter Cc: r-help@stat.math.ethz.ch Subject: Re: [R] thousand separator (was RE: weight) Bert, What am I missing? print(as.numeric(gsub(,, , 1,123,456.789)), 10) [1] 1123456.789 FWIW, this is using: R version 2.5.0 Patched (2007-04-27 r41355) Marc On Mon, 2007-04-30 at 10:13 -0700, Bert Gunter wrote: Except this doesn't work for 1,123,456.789 Marc. I hesitate to suggest it, but gregexpr() will do it, as it captures the position of **every** match to ,. This could be then used to process the vector via some sort of loop/apply statement. But I think there **must** be a more elegant way using regular expressions alone, so I, too, await a clever reply. -- Bert Bert Gunter Genentech Nonclinical Statistics -Original Message- From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED] On Behalf Of Marc Schwartz Sent: Monday, April 30, 2007 10:02 AM To: Liaw, Andy Cc: r-help@stat.math.ethz.ch Subject: Re: [R] thousand separator (was RE: weight) One possibility would be to use something like the following post-import: WTPP [1] 1,106.8250 1,336.5138 str(WTPP) Factor w/ 2 levels 1,106.8250,1,336.5138: 1 2 as.numeric(gsub(,, , WTPP)) [1] 1106.825 1336.514 Essentially strip the ',' characters from the factors and then coerce the resultant character vector to numeric. HTH, Marc Schwartz On Mon, 2007-04-30 at 12:26 -0400, Liaw, Andy wrote: I've run into this occasionally. My current solution is simply to read it into Excel, re-format the offending column(s) by unchecking the thousand separator box, and write it back out. Not exactly ideal to say the least. If anyone can provide a better solution in R, I'm all ears... Andy From: Natalie O'Toole Hi, These are the variables in my file. I think the variable i'm having problems with is WTPP which is of the Factor type. Does anyone know how to fix this, please? Thanks, Nat data.frame': 290 obs. of 5 variables: $ PROV : num 48 48 48 48 48 48 48 48 48 48 ... $ REGION: num 4 4 4 4 4 4 4 4 4 4 ... $ GRADE : num 7 7 7 7 7 7 7 7 7 7 ... $ Y_Q10A: num 1.1 1.1 1.1 1.1 1.1 1.1 1.1 1.1 1.1 1.1 ... $ WTPP : Factor w/ 1884 levels 1,106.8250,1,336.5138,..: 1544 67 1568 40 221 1702 1702 1434 310 310 ... __ --- Douglas Bates [EMAIL PROTECTED] wrote: On 4/28/07, John Kane [EMAIL PROTECTED] wrote: === message truncated === __ R-help@stat.math.ethz.ch 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] Problem with the installation of install R on Sun Solaris
Greetings. It looks like your R config can't find your C compiler. Do you have / usr/ccs/bin in your path? Or, if you installed Studio 11, do you have /usr/ccs/bin and /opt/SUNWspro/bin in your path? Hope this helps, Jon __ R-help@stat.math.ethz.ch 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-help@stat.math.ethz.ch 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] Problem with the installation of install R on Sun Solaris
Vipin Singhal wrote: (nothing, except for an attached file) configure:3749: checking for gcc configure:3778: result: no configure:3829: checking for cc configure:3858: result: no configure:3871: checking for cc configure:3917: result: no configure:3970: checking for cl configure:3999: result: no configure:4013: error: no acceptable C compiler found in $PATH See `config.log' for more details. Which part of this is it that you have trouble understanding? __ R-help@stat.math.ethz.ch 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] to draw a smooth arc
This thread prompts me to ask about something I've been pondering for a while, as to whether there's an implementation somewhere ticked away in the R resources. So far, people have been responding to the original query in terms of increasing the numbers of points, and joining these by lines. However, if you're using PostScript output, you can draw really smooth curves by exploiting PS's curveto operator. This draws a cubic-curve segment in the following way: The two points you want to join with a curve will be denoted by (X0,Y0) and (X3,Y3) in the following (for reasons which will appear). The PS command is of the form x1 y1 x2 y2 X3 Y3 curevto At (X0,Y0) the tangent to the curve (as it departs from (X0,Y0) is in the direction of the directed line from (X0,Y0) to (x1,y1), and at (X3,Y3) (as it arrives) the tangent to the curve is in the direction of the directed line from (x2,y3) to (X3,Y3). The location of (X0,Y0) is not part of the command, since it is implicit in the PS currentpoint which is the starting point of the curve. The result is (in theory, and in practice to within the resolution of the output device) a perfectly smooth curve, provided the consecutive cubic segments have the same tangent at each of the points being joined. This can be achieved by appropriate choice of the intermediate points -- (x1,y2), (x2,y2) above. So far, when I've done this myself (including when using the output from R to give the points being joined), I've done the computation of the intermediate points by hand. This basically involves deciding, at each of the points being joined, what the tangent to the smooth curve shouold be. Of course, there is an element of arbitrariness in this, unless there is an analytic representation of the curve on which the points lie (e.g. you're plotting sin(x)/x every pi/8, and want to join them smoothly), when all you need is the derivatives at the points. Crudely, you might evaluate the direction at a point in terms os a weighted average of the directions to its two immediate neighbours (the nearer meghbour ges the greater weight); less crudely, you might fit a quadratic through the point and its 2 neighbours and use the gradient at the middle point; and so on. Once you've decided on the tangent at each point, it's then straightforward to compute suitable intermediate points to serve as (x1,y2) and (x2,y2). (One application where this sort of approach is needed is in joining computed points on iso-contours, where the individual points have been determined by interpolation of spot-measurements at nearby measuring stations). Anyway. The Question: is there a general function for the above kind of smooth curve-drawing? With thanks, Ted. E-Mail: (Ted Harding) [EMAIL PROTECTED] Fax-to-email: +44 (0)870 094 0861 Date: 01-May-07 Time: 14:50:38 -- XFMail -- __ R-help@stat.math.ethz.ch 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] Polar graph of time and tide
I have been trying to visualize times of lowest tides, month by month. I have tide predictions with times either in unix time or a text format, and heights in feet or meters. I had been able to derive the clock times of each prediction. I would now like to graph this data with points showing heights as r and times as theta, from to 2355. There is a seasonal component: I am interested in displaying times of lowest tides in particular. I am sure this is so simple as to burden those on the list; I however have spent two evenings trying to figure out how to use polar.plot, and I'm not sure that's the best way to do this. May I request some advice? The docs with polar.plot are not complete, I fear. Thank you, begging for your indulgence, Alan -- Alan Davis, Kagman High School, Saipan [EMAIL PROTECTED] Every great advance in natural knowledge has involved the absolute rejection of authority. - Thomas H. Huxley __ R-help@stat.math.ethz.ch 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] to draw a smooth arc
There is the grid.xspline function in the grid package that allows for things like this (the control points, though more general than what you state). I don't know if it uses the postscript curveto, or approximates using line segments. You can also use the xfig device, then use xfig, winfig, or jfig to explicitly convert any polylines to xslpines, adjust any parameters of the spline that you want, then export to other formats. Hope this helps, -- Gregory (Greg) L. Snow Ph.D. Statistical Data Center Intermountain Healthcare [EMAIL PROTECTED] (801) 408-8111 -Original Message- From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED] On Behalf Of [EMAIL PROTECTED] Sent: Tuesday, May 01, 2007 7:51 AM To: r-help@stat.math.ethz.ch Subject: Re: [R] to draw a smooth arc This thread prompts me to ask about something I've been pondering for a while, as to whether there's an implementation somewhere ticked away in the R resources. So far, people have been responding to the original query in terms of increasing the numbers of points, and joining these by lines. However, if you're using PostScript output, you can draw really smooth curves by exploiting PS's curveto operator. This draws a cubic-curve segment in the following way: The two points you want to join with a curve will be denoted by (X0,Y0) and (X3,Y3) in the following (for reasons which will appear). The PS command is of the form x1 y1 x2 y2 X3 Y3 curevto At (X0,Y0) the tangent to the curve (as it departs from (X0,Y0) is in the direction of the directed line from (X0,Y0) to (x1,y1), and at (X3,Y3) (as it arrives) the tangent to the curve is in the direction of the directed line from (x2,y3) to (X3,Y3). The location of (X0,Y0) is not part of the command, since it is implicit in the PS currentpoint which is the starting point of the curve. The result is (in theory, and in practice to within the resolution of the output device) a perfectly smooth curve, provided the consecutive cubic segments have the same tangent at each of the points being joined. This can be achieved by appropriate choice of the intermediate points -- (x1,y2), (x2,y2) above. So far, when I've done this myself (including when using the output from R to give the points being joined), I've done the computation of the intermediate points by hand. This basically involves deciding, at each of the points being joined, what the tangent to the smooth curve shouold be. Of course, there is an element of arbitrariness in this, unless there is an analytic representation of the curve on which the points lie (e.g. you're plotting sin(x)/x every pi/8, and want to join them smoothly), when all you need is the derivatives at the points. Crudely, you might evaluate the direction at a point in terms os a weighted average of the directions to its two immediate neighbours (the nearer meghbour ges the greater weight); less crudely, you might fit a quadratic through the point and its 2 neighbours and use the gradient at the middle point; and so on. Once you've decided on the tangent at each point, it's then straightforward to compute suitable intermediate points to serve as (x1,y2) and (x2,y2). (One application where this sort of approach is needed is in joining computed points on iso-contours, where the individual points have been determined by interpolation of spot-measurements at nearby measuring stations). Anyway. The Question: is there a general function for the above kind of smooth curve-drawing? With thanks, Ted. E-Mail: (Ted Harding) [EMAIL PROTECTED] Fax-to-email: +44 (0)870 094 0861 Date: 01-May-07 Time: 14:50:38 -- XFMail -- __ R-help@stat.math.ethz.ch 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-help@stat.math.ethz.ch 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] Polar graph of time and tide
Alan E. Davis [EMAIL PROTECTED] wrote in message news:[EMAIL PROTECTED] I have been trying to visualize times of lowest tides, month by month. I have tide predictions with times either in unix time or a text format, and heights in feet or meters. I had been able to derive the clock times of each prediction. I would now like to graph this data with points showing heights as r and times as theta, from to 2355. There is a seasonal component: I am interested in displaying times of lowest tides in particular. Does this get you started? library(plotrix) theta - seq(0, 23.5, by=0.5) r - runif(length(theta), 5, 10) clock24.plot(r, theta, main=Polar Plot) or clock24.plot(r, theta, main=Polar Plot, rp.type=p) efg Stowers Institute for Medical Research __ R-help@stat.math.ethz.ch 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] creating eps files
Hello, For a long time, I have been creating eps files from R using the following command: dev.copy2eps(file=my.eps) This has worked very well. But recently, the compositor of a journal is complaining that The eps files would be useable except that they have not converted the type to outlines Sorry for being vague, but I have no idea what this compositor is talking about. Do people have similar experience? And is there an option in the dev.copy2eps() that I could use to fix this problem? Many thanks for any suggestion. Best, Wing dept addr: Department of Sociology, University of Oxford, Manor Road, Oxford OX1 3UQ, UK college addr: New College, Holywell Street, Oxford OX1 3BN, UK dept tel: +44 (1865) 286176; college tel: +44 (1865) 279593 http://users.ox.ac.uk/~sfos0006 __ R-help@stat.math.ethz.ch 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] to draw a smooth arc
Here is an approach that clips the circle you like from symbols down to an arc (this will work as long as the arc is less than half a circle, for arcs greater than half a circle, you could draw the whole circle then use this to draw an arc of the bacground color over the section you don't want): library(TeachingDemos) plot(-5:5, -5:5, type='n') clipplot( symbols(0,0,circles=2, add=TRUE), c(0,5), c(0,5) ) Hope this helps, -- Gregory (Greg) L. Snow Ph.D. Statistical Data Center Intermountain Healthcare [EMAIL PROTECTED] (801) 408-8111 -Original Message- From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED] On Behalf Of Paulo Barata Sent: Monday, April 30, 2007 8:17 PM To: r-help@stat.math.ethz.ch Subject: [R] to draw a smooth arc Dear R-list members, I would like to draw a smooth arc. I can draw an arc parametrically, but this produces an arc too coarse, even allowing for different increments in sequence t in the example below. Function symbols (graphics) does produce a smooth circle, but it cannot produce an arc. Please see the following example, drawing complete circles: plot(-5:5,-5:5,type='n') ## draws circle with function symbols (package graphics) ## - inner circle is very smooth: symbols(0,0,circles=2,add=TRUE) ## draws circle parametrically - outer circle is too coarse: pi - 4*atan(1) t - seq(0,2*pi,0.02) lines(4*cos(t),4*sin(t)) Package plotrix has a function draw.arc, but arcs produced with this function are also either too coarse or too polygonal, depending on the number of polygons used to approximate the arc. Is there a way to harness the characteristics of function symbols (graphics) to draw a smooth arc, not just a complete circle? I am using R 2.5.0, running under Windows XP. Thank you very much. Paulo Barata - Paulo Barata Fundacao Oswaldo Cruz (Oswaldo Cruz Foundation) Rua Leopoldo Bulhoes 1480 - 8A 21041-210 Rio de Janeiro - RJ Brazil E-mail: [EMAIL PROTECTED] __ R-help@stat.math.ethz.ch 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-help@stat.math.ethz.ch 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] creating eps files
On Tue, 1 May 2007, Tak Wing Chan wrote: Hello, For a long time, I have been creating eps files from R using the following command: dev.copy2eps(file=my.eps) This has worked very well. But recently, the compositor of a journal is complaining that The eps files would be useable except that they have not converted the type to outlines Sorry for being vague, but I have no idea what this compositor is talking about. Do people have similar experience? And is there an option in the dev.copy2eps() that I could use to fix this problem? I suspect he means that he wants the fonts embedded, which is not what eps requires. See the article in R-News about doing that: http://cran.r-project.org/doc/Rnews/Rnews_2006-2.pdf -- 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 __ R-help@stat.math.ethz.ch 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] Concepts question: environment, frame, search path
On Tue, 1 May 2007, Duncan Murdoch wrote: On 01/05/2007 12:29 AM, Graham Wideman wrote: Folks: I'd appreciate if someone could straighten me out on a few concepts which are described a bit ambiguously in the docs. 1. data.frame: Refan p84: 'A data frame is a list of variables of the same length with unique row names, given class data.frame.' I probably don't need to point out how opaque that is! Which manual are you looking at? The reference index (refman.pdf)? It doesn't usually include statements like that; they are usually found in the Introduction to R (R-intro.pdf) or the R Language Definition (R-lang.pdf). But since the refman is just a collection of man pages, it might be in there somewhere. And since the manuals do get updated, that statement may not be present in the current release. (I did a quick search of the source, and couldn't spot it, but my search might have failed because of line breaks, strange formatting, or looking in the wrong place.) By the way, it's generally best to cite the section name where you found a quote, because the pagination varies from system to system. Even better would be to give a URL to the online HTML version at http://cran.r-project.org/manuals.html. For future reference, if you are suggesting a change, it's best to cite the line number in the source at https://svn.r-project.org/R/trunk/doc/manual in the *.texi files or https://svn.r-project.org/R/trunk/src/library/*/man/*.Rd for man pages, and send such suggestions to the R-devel list. Anyhow, key question: Some places in the docs seem pretty firm that a data.frame is basically a 2-D array with: a) named rows and b) columns whose items within a column be of uniform data type. Elsewhere, it seems like a data.frame can be a collection of arbitrary variables. The former interpretation is correct. Since the variables all have the same length, things like df[i, j] make sense: they choose the i'th entry from the j'th variable (according to the refan definition), or the i'th row, j'th column (according to the 2-D array interpretation. 2. environment --- Refman p122: Environments consist of a frame, or collection of named objects, and a pointer to an enclosing environment. Is the or here explaining parenthetically that a frame is a collection of named objects, or is separating too alternative structures for an environment? The former. If the former, does this imply that a frame can contain arbitrary variables? Yes, but a frame isn't an R object, it's a concept that appears in descriptions, e.g. part of an environment, or the local variables created during function evaluation, etc. And pointer? Is that a type of thing in R? No, there are no pointers in R. There are a couple of tricks to fake them (e.g. environment objects aren't copied when assigned, you just get a new reference to the same environment; this allows you to construct something like a pointer by wrapping an object in an environment), but I don't recommend using these routinely. Nevertheless, the statement is true. R is implemented using pointers. 3. R search path; attach() The R search path appears to hold the list of collections of data (my term) that can be accessed by a users' commands. Refman p27 tells that search path can hold items that are data.frame, list, environment or R data file (on disk). Yet R-intro p28 describes attach() as taking a directory name argument. What is the concept directory in this context? I haven't read the preceding pages carefully, but that looks like an error. The usual argument to attach is a package name, and what gets attached is an environment holding the exports from the package. Packages are stored in directories in the file system, so maybe that's what the author of that line had in mind. For the record, it is old S terminology: that document was converted from notes for S. What S(-PLUS) now calls 'chapters' it used to call directories. Also for the record, these documents do not have page numbers: their layout depends on the version of R, paper size and the tools used to prepare them. -- 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 __ R-help@stat.math.ethz.ch 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] logrank test
how do l programme the logrank test. l am trying to compare 2 survival curves - [[alternative HTML version deleted]] __ R-help@stat.math.ethz.ch 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] logrank test
raymond chiruka wrote: how do l programme the logrank test. l am trying to compare 2 survival curves library(survival) ?survdiff __ R-help@stat.math.ethz.ch 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] logrank test
Hi, On 5/1/07, raymond chiruka [EMAIL PROTECTED] wrote: how do l programme the logrank test. l am trying to compare 2 survival curves if you simply want to use the logrank test, have a look at the first example of the function survdiff in the survival package. If you read the help page there, it says that the default setting of rho=0 is the log rank test. library(survival) survdiff(Surv(futime, fustat) ~ rx,data=ovarian) survdiff(Surv(futime, fustat) ~ rx,data=ovarian, rho=0) I hope this helps? Best, Roland [[alternative HTML version deleted]] __ R-help@stat.math.ethz.ch 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] logrank test
On Tue, 2007-05-01 at 08:34 -0700, raymond chiruka wrote: how do l programme the logrank test. l am trying to compare 2 survival curves See: library(survival) ?survdiff and take note of the 'rho' argument, which when set to 0 is the logrank test. HTH, Marc Schwartz __ R-help@stat.math.ethz.ch 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] linout=TRUE in nnet package ?
Hello, I am trying to figure out what nnet does when you select nnet =TRUE, I understand that it provides you with linear outputs, but I don't understand how it arrives at those linear outputs. I assume that nnet still applies the logistic function as the activation function for the nodes in the hidden layer, but I also assume another activation function must be applied at the output layer in order to return the predictions to the linear scale. I am not able to figure out what is happening from the documentation, could anyone provide more details about what nnet does at the output layer? best, Spencer [[alternative HTML version deleted]] __ R-help@stat.math.ethz.ch 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] to draw a smooth arc
On Tue, 1 May 2007, Greg Snow wrote: Here is an approach that clips the circle you like from symbols down to an arc (this will work as long as the arc is less than half a circle, for arcs greater than half a circle, you could draw the whole circle then use this to draw an arc of the bacground color over the section you don't want): library(TeachingDemos) plot(-5:5, -5:5, type='n') clipplot( symbols(0,0,circles=2, add=TRUE), c(0,5), c(0,5) ) I had considered this approach: clipping a circle to a rectangle isn't strictly an arc, as will be clear if the line width is large. Consider clipplot(symbols(0, 0 ,circles=2, add=TRUE, lwd=5), c(-1,5), c(-1,5)) Note too that what happens with clipping is device-dependent. If R's internal clipping is used, the part-circle is converted to a polygon. -- 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 __ R-help@stat.math.ethz.ch 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] logrank test
And since Peter Dalgaard also just answered (without advertising his book): if you (or your library) happen to have 'Introductory Statistics with R' by Peter Dalgaard, have a look at section 12.4. Best, Roland On 5/1/07, raymond chiruka [EMAIL PROTECTED] wrote: how do l programme the logrank test. l am trying to compare 2 survival curves - [[alternative HTML version deleted]] __ R-help@stat.math.ethz.ch 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. [[alternative HTML version deleted]] __ R-help@stat.math.ethz.ch 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] Matrix column name
Dear R users, Having searched the mail archive I think the conclusion was that it is not possible to have a column name when there is only one column in the matrix. But I thought I'd check with the more experienced users. What I tried to do was: in a loop I pick a column, record the column name and remove the column from the matrix. But when there were 2 columns left, after one column was removed, the last column name disappeared by default. It means that I always miss out the last column. I tried this by hand: matrix.a 801 802 803 [1,] -0.0906346 0.0906346 0.0906346 [2,] -0.0804911 0.0804911 0.0804911 [3,] -0.0703796 0.0703796 0.0703796 matrix.a-as.matrix(matrix.a[,-1]) matrix.a 802 803 [1,] 0.0906346 0.0906346 [2,] 0.0804911 0.0804911 [3,] 0.0703796 0.0703796 matrix.a-as.matrix(matrix.a[,-1]) matrix.a [,1] [1,] 0.0906346 [2,] 0.0804911 [3,] 0.0703796 Is there a way to force the column name to remain in such a case? Thanks, Alex sessionInfo() R version 2.4.1 (2006-12-18) i386-pc-mingw32 locale: LC_COLLATE=English_United Kingdom.1252;LC_CTYPE=English_United Kingdom.1252;LC_MONETARY=English_United Kingdom.1252;LC_NUMERIC=C;LC_TIME=English_United Kingdom.1252 attached base packages: [1] stats graphics grDevices utils datasets methods [7] base Alex Lam PhD student Department of Genetics and Genomics Roslin Institute (Edinburgh) Roslin Midlothian EH25 9PS Great Britain Phone +44 131 5274471 Web http://www.roslin.ac.uk Roslin Institute is a company limited by guarantee, registered in Scotland (registered number SC157100) and a Scottish Charity (registered number SC023592). Our registered office is at Roslin, Midlothian, EH25 9PS. VAT registration number 847380013. The information contained in this e-mail (including any attachments) is confidential and is intended for the use of the addressee only. The opinions expressed within this e-mail (including any attachments) are the opinions of the sender and do not necessarily constitute those of Roslin Institute (Edinburgh) (the Institute) unless specifically stated by a sender who is duly authorised to do so on behalf of the Institute __ R-help@stat.math.ethz.ch 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] Simulation using parts of density function
Hi My simulation with the followin R code works perfectly: sim - replicate(999, sum(exp(rgamma(rpois(1,2000), scale = 0.5, shape = 12 But now I do not want to have values in object sim exceeding 5'000'000, that means that I am just using the beginning of densitiy function gamma x 15.4. Is there a possibility to modify my code in an easy way? Thanks for any help! Regards, Brigitte [[alternative HTML version deleted]] __ R-help@stat.math.ethz.ch 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] Matrix column name
On Tue, 2007-05-01 at 18:03 +0100, alex lam (RI) wrote: Dear R users, Having searched the mail archive I think the conclusion was that it is not possible to have a column name when there is only one column in the matrix. But I thought I'd check with the more experienced users. What I tried to do was: in a loop I pick a column, record the column name and remove the column from the matrix. But when there were 2 columns left, after one column was removed, the last column name disappeared by default. It means that I always miss out the last column. See R FAQ 7.5 Why do my matrices lose dimensions: http://cran.r-project.org/doc/FAQ/R-FAQ.html#Why-do-my-matrices-lose-dimensions_003f which has some examples, along with ?Extract To wit: MAT - matrix(1:12, ncol = 3) colnames(MAT) - LETTERS[1:3] MAT A B C [1,] 1 5 9 [2,] 2 6 10 [3,] 3 7 11 [4,] 4 8 12 MAT[, 1] [1] 1 2 3 4 MAT[, 1, drop = FALSE] A [1,] 1 [2,] 2 [3,] 3 [4,] 4 HTH, Marc Schwartz __ R-help@stat.math.ethz.ch 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] calculating area under the ROC curve
Hi, I recently did a study where we gathered decisions and confidence ratings. My understanding is that I can convert this to a ROC curve by getting hits and false alarms at the various confidence ratings. I figured out that part of the problem. I noticed a few functions for calculating AUC. Are there any preferred ones for this particular kind of design? __ R-help@stat.math.ethz.ch 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] to draw a smooth arc
Dr. Snow and Prof. Ripley, Dr. Snow's suggestion, using clipplot (package TeachingDemos), is maybe a partial solution to the problem of drawing an arc of a circle (as long as the line width of the arc is not that large, as pointed out by Prof. Ripley). If the arc is symmetrical around a vertical line, then it is not so difficult to draw it that way. But an arc that does not have this kind of symmetry would possibly require some geometrical computations to find the proper rectangle to be used for clipping. I would like to suggest that in a future version of R some function be included in the graphics package to draw smooth arcs with given center, radius, initial and final angles. I suppose that the basic ingredients are available in function symbols (graphics). Thank you very much. Paulo Barata (Rio de Janeiro - Brazil) --- Prof Brian Ripley wrote: On Tue, 1 May 2007, Greg Snow wrote: Here is an approach that clips the circle you like from symbols down to an arc (this will work as long as the arc is less than half a circle, for arcs greater than half a circle, you could draw the whole circle then use this to draw an arc of the bacground color over the section you don't want): library(TeachingDemos) plot(-5:5, -5:5, type='n') clipplot( symbols(0,0,circles=2, add=TRUE), c(0,5), c(0,5) ) I had considered this approach: clipping a circle to a rectangle isn't strictly an arc, as will be clear if the line width is large. Consider clipplot(symbols(0, 0 ,circles=2, add=TRUE, lwd=5), c(-1,5), c(-1,5)) Note too that what happens with clipping is device-dependent. If R's internal clipping is used, the part-circle is converted to a polygon. __ R-help@stat.math.ethz.ch 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] sorting in barplot
Hello, I'm trying to sort my bargraph.CI plot (function like barplot in the SCIPLOT package) by the mean of the response variable. Does somebody have a trick for it. Thank you. Romain Mayor, PHD student. __ R-help@stat.math.ethz.ch 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] sorting in barplot
On Tue, 2007-05-01 at 19:33 +0200, [EMAIL PROTECTED] wrote: Hello, I'm trying to sort my bargraph.CI plot (function like barplot in the SCIPLOT package) by the mean of the response variable. Does somebody have a trick for it. Thank you. Romain Mayor, PHD student. If it is built on top of barplot(), then by default, the factor levels of your response variable will determine the order of the bars in the plot. See ?reorder.factor for more details relative to defining the order based upon the mean of the variable. There is an example there of using the median. HTH, Marc Schwartz __ R-help@stat.math.ethz.ch 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] Simulation using parts of density function
On 01-May-07 17:03:46, Thür Brigitte wrote: Hi My simulation with the followin R code works perfectly: sim - replicate(999, sum(exp(rgamma(rpois(1,2000), scale = 0.5, shape = 12 But now I do not want to have values in object sim exceeding 5'000'000, that means that I am just using the beginning of densitiy function gamma x 15.4. Is there a possibility to modify my code in an easy way? Thanks for any help! Regards, Brigitte A somewhat extreme problem! The easiest way to modify the code is as below -- certiainly easier than writing a special function to draw random samples from the truncated gamma distribution. A bit of experimentation shows that, from your code above, about 10% of the results are = 500. So: sim-NULL remain - 999 while(remain0){ sim0-replicate(10*remain, sum(exp(rgamma(rpois(1,2000), scale = 0.5, shape = 12))) ) sim-c(sim,sim0[sim0=500]) remain-(999 - length(sim)) } sim-sim[1:999] Results of a run: sum(sim500) [1] 0 max(sim) [1] 4999696 length(sim) [1] 999 It may be on the slow side (though not hugely -- on a quite slow machine the above run was completed in 2min 5sec, while the 999-replicate in your original took 15sec. So about 8 times as long. Most of this, of course, is taken up with the first round. Hoping this helps, Ted. E-Mail: (Ted Harding) [EMAIL PROTECTED] Fax-to-email: +44 (0)870 094 0861 Date: 01-May-07 Time: 19:18:01 -- XFMail -- __ R-help@stat.math.ethz.ch 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] Free Webinar: Vendor Neutral Intro to Data Mining for Absolute Beginners, May 23, 2007
ONLINE VENDOR NEUTRAL INTRO TO DATA MINING FOR ABSOLUTE BEGINNERS (no charge) A non-technical data mining introduction for absolute beginners May 23, 2007, 10AM - 11AM PST Future Sessions (June 14, Sept 7) To register for the webinar --- 1. Go to https://salford.webex.com/salford/onstage/g.php?d=928318845t=a 2. Click Enroll. 3. On the registration form, enter your information and then click Submit. Once you have registered, you will receive a confirmation email message with instructions on how to join the event, as well as audio and system requirements. Please read this confirmation email carefully! This one-hour webinar is a perfect place to start if you are new to data mining and have little-to-no background in statistics or machine learning. In one hour, we will discuss: **Data basics: what kind of data is required for data mining and predictive analytics; In what format must the data be; what steps are necessary to prepare data appropriately **What kinds of questions can we answer with data mining **How data mining models work: the inputs, the outputs, and the nature of the predictive mechanism **Evaluation criteria: how predictive models can be assessed and their value measured **Specific background knowledge to prepare you to begin a data mining project. Please do not hesitate to contact me if you have any questions. Sincerely, Lisa Solomon [EMAIL PROTECTED] __ R-help@stat.math.ethz.ch 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] to draw a smooth arc
Hi Paulo Barata wrote: Dr. Snow and Prof. Ripley, Dr. Snow's suggestion, using clipplot (package TeachingDemos), is maybe a partial solution to the problem of drawing an arc of a circle (as long as the line width of the arc is not that large, as pointed out by Prof. Ripley). If the arc is symmetrical around a vertical line, then it is not so difficult to draw it that way. But an arc that does not have this kind of symmetry would possibly require some geometrical computations to find the proper rectangle to be used for clipping. I would like to suggest that in a future version of R some function be included in the graphics package to draw smooth arcs with given center, radius, initial and final angles. I suppose that the basic ingredients are available in function symbols (graphics). Just to back up a few previous posts ... There is something like this facility already available via the grid.xspline() function in the grid package. This provides very flexible curve drawing (including curves very close to Bezier curves) based on the X-Splines implemented in xfig. The grid.curve() function provides a convenience layer that allows for at least certain parameterisations of arcs (you specify the arc end points and the angle). These functions are built on functionality within the core graphics engine, so exposing a similar interface (e.g., an xspline() function) within traditional graphics would be relatively straightforward. The core functionality draws the curves as line segments (but automatically figures out how many segments to use so that the curve looks smooth); it does NOT call curve-drawing primitives in the graphics device (like PostScript's curveto). In summary: there is some support for smooth curves, but we could still benefit from a specific arc() function with the standard centre-radius-angle parameterisation and we could also benefit from exposing the native strengths of different graphics devices (rather than the current lowest-common-denominator approach). Paul Thank you very much. Paulo Barata (Rio de Janeiro - Brazil) --- Prof Brian Ripley wrote: On Tue, 1 May 2007, Greg Snow wrote: Here is an approach that clips the circle you like from symbols down to an arc (this will work as long as the arc is less than half a circle, for arcs greater than half a circle, you could draw the whole circle then use this to draw an arc of the bacground color over the section you don't want): library(TeachingDemos) plot(-5:5, -5:5, type='n') clipplot( symbols(0,0,circles=2, add=TRUE), c(0,5), c(0,5) ) I had considered this approach: clipping a circle to a rectangle isn't strictly an arc, as will be clear if the line width is large. Consider clipplot(symbols(0, 0 ,circles=2, add=TRUE, lwd=5), c(-1,5), c(-1,5)) Note too that what happens with clipping is device-dependent. If R's internal clipping is used, the part-circle is converted to a polygon. __ R-help@stat.math.ethz.ch 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. -- Dr Paul Murrell Department of Statistics The University of Auckland Private Bag 92019 Auckland New Zealand 64 9 3737599 x85392 [EMAIL PROTECTED] http://www.stat.auckland.ac.nz/~paul/ __ R-help@stat.math.ethz.ch 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] Matrix column name
You seem to be looking for matrix.a[,-1, drop = TRUE] On Tue, 1 May 2007, alex lam (RI) wrote: Dear R users, Having searched the mail archive I think the conclusion was that it is not possible to have a column name when there is only one column in the matrix. But I thought I'd check with the more experienced users. What I tried to do was: in a loop I pick a column, record the column name and remove the column from the matrix. But when there were 2 columns left, after one column was removed, the last column name disappeared by default. It means that I always miss out the last column. And the matrix became a vector. I tried this by hand: matrix.a 801 802 803 [1,] -0.0906346 0.0906346 0.0906346 [2,] -0.0804911 0.0804911 0.0804911 [3,] -0.0703796 0.0703796 0.0703796 matrix.a-as.matrix(matrix.a[,-1]) matrix.a 802 803 [1,] 0.0906346 0.0906346 [2,] 0.0804911 0.0804911 [3,] 0.0703796 0.0703796 matrix.a-as.matrix(matrix.a[,-1]) matrix.a [,1] [1,] 0.0906346 [2,] 0.0804911 [3,] 0.0703796 Is there a way to force the column name to remain in such a case? Thanks, Alex sessionInfo() R version 2.4.1 (2006-12-18) i386-pc-mingw32 locale: LC_COLLATE=English_United Kingdom.1252;LC_CTYPE=English_United Kingdom.1252;LC_MONETARY=English_United Kingdom.1252;LC_NUMERIC=C;LC_TIME=English_United Kingdom.1252 attached base packages: [1] stats graphics grDevices utils datasets methods [7] base Alex Lam PhD student Department of Genetics and Genomics Roslin Institute (Edinburgh) Roslin Midlothian EH25 9PS Great Britain Phone +44 131 5274471 Web http://www.roslin.ac.uk -- 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 __ R-help@stat.math.ethz.ch 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] to calculate sums
Dear R-users, I am trying to use R to calculate sums like the ones in the file attached. Would you please provide some help? At the moment I have no clue about how to due... Thank you in advance, Kind regards, Pedro __ R-help@stat.math.ethz.ch 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] dlda{supclust} 's output
Hi, I am using dlda algorithm from supclust package and I am wondering if the output can be a continuous probability instead of discrete class label (zero or one) since it puts some restriction on convariance matrix, compared with lda, while the latter can. thanks, -- Weiwei Shi, Ph.D Research Scientist GeneGO, Inc. Did you always know? No, I did not. But I believed... ---Matrix III __ R-help@stat.math.ethz.ch 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] to draw a smooth arc
Dr. Murrell and all, One final suggestion: a future function arc() in package graphics, with centre-radius-angle parameterisation, could also include an option to draw arrows at either end of the arc, as one can find in function arrows(). Thank you. Paulo Barata --- Paul Murrell wrote: Hi Paulo Barata wrote: Dr. Snow and Prof. Ripley, Dr. Snow's suggestion, using clipplot (package TeachingDemos), is maybe a partial solution to the problem of drawing an arc of a circle (as long as the line width of the arc is not that large, as pointed out by Prof. Ripley). If the arc is symmetrical around a vertical line, then it is not so difficult to draw it that way. But an arc that does not have this kind of symmetry would possibly require some geometrical computations to find the proper rectangle to be used for clipping. I would like to suggest that in a future version of R some function be included in the graphics package to draw smooth arcs with given center, radius, initial and final angles. I suppose that the basic ingredients are available in function symbols (graphics). Just to back up a few previous posts ... There is something like this facility already available via the grid.xspline() function in the grid package. This provides very flexible curve drawing (including curves very close to Bezier curves) based on the X-Splines implemented in xfig. The grid.curve() function provides a convenience layer that allows for at least certain parameterisations of arcs (you specify the arc end points and the angle). These functions are built on functionality within the core graphics engine, so exposing a similar interface (e.g., an xspline() function) within traditional graphics would be relatively straightforward. The core functionality draws the curves as line segments (but automatically figures out how many segments to use so that the curve looks smooth); it does NOT call curve-drawing primitives in the graphics device (like PostScript's curveto). In summary: there is some support for smooth curves, but we could still benefit from a specific arc() function with the standard centre-radius-angle parameterisation and we could also benefit from exposing the native strengths of different graphics devices (rather than the current lowest-common-denominator approach). Paul Thank you very much. Paulo Barata (Rio de Janeiro - Brazil) --- Prof Brian Ripley wrote: On Tue, 1 May 2007, Greg Snow wrote: Here is an approach that clips the circle you like from symbols down to an arc (this will work as long as the arc is less than half a circle, for arcs greater than half a circle, you could draw the whole circle then use this to draw an arc of the bacground color over the section you don't want): library(TeachingDemos) plot(-5:5, -5:5, type='n') clipplot( symbols(0,0,circles=2, add=TRUE), c(0,5), c(0,5) ) I had considered this approach: clipping a circle to a rectangle isn't strictly an arc, as will be clear if the line width is large. Consider clipplot(symbols(0, 0 ,circles=2, add=TRUE, lwd=5), c(-1,5), c(-1,5)) Note too that what happens with clipping is device-dependent. If R's internal clipping is used, the part-circle is converted to a polygon. __ R-help@stat.math.ethz.ch 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-help@stat.math.ethz.ch 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] to draw a smooth arc
Hi Paulo Barata wrote: Dr. Murrell and all, One final suggestion: a future function arc() in package graphics, with centre-radius-angle parameterisation, could also include an option to draw arrows at either end of the arc, as one can find in function arrows(). ... and in grid.xspline() and grid.curve(). Paul Thank you. Paulo Barata --- Paul Murrell wrote: Hi Paulo Barata wrote: Dr. Snow and Prof. Ripley, Dr. Snow's suggestion, using clipplot (package TeachingDemos), is maybe a partial solution to the problem of drawing an arc of a circle (as long as the line width of the arc is not that large, as pointed out by Prof. Ripley). If the arc is symmetrical around a vertical line, then it is not so difficult to draw it that way. But an arc that does not have this kind of symmetry would possibly require some geometrical computations to find the proper rectangle to be used for clipping. I would like to suggest that in a future version of R some function be included in the graphics package to draw smooth arcs with given center, radius, initial and final angles. I suppose that the basic ingredients are available in function symbols (graphics). Just to back up a few previous posts ... There is something like this facility already available via the grid.xspline() function in the grid package. This provides very flexible curve drawing (including curves very close to Bezier curves) based on the X-Splines implemented in xfig. The grid.curve() function provides a convenience layer that allows for at least certain parameterisations of arcs (you specify the arc end points and the angle). These functions are built on functionality within the core graphics engine, so exposing a similar interface (e.g., an xspline() function) within traditional graphics would be relatively straightforward. The core functionality draws the curves as line segments (but automatically figures out how many segments to use so that the curve looks smooth); it does NOT call curve-drawing primitives in the graphics device (like PostScript's curveto). In summary: there is some support for smooth curves, but we could still benefit from a specific arc() function with the standard centre-radius-angle parameterisation and we could also benefit from exposing the native strengths of different graphics devices (rather than the current lowest-common-denominator approach). Paul Thank you very much. Paulo Barata (Rio de Janeiro - Brazil) --- Prof Brian Ripley wrote: On Tue, 1 May 2007, Greg Snow wrote: Here is an approach that clips the circle you like from symbols down to an arc (this will work as long as the arc is less than half a circle, for arcs greater than half a circle, you could draw the whole circle then use this to draw an arc of the bacground color over the section you don't want): library(TeachingDemos) plot(-5:5, -5:5, type='n') clipplot( symbols(0,0,circles=2, add=TRUE), c(0,5), c(0,5) ) I had considered this approach: clipping a circle to a rectangle isn't strictly an arc, as will be clear if the line width is large. Consider clipplot(symbols(0, 0 ,circles=2, add=TRUE, lwd=5), c(-1,5), c(-1,5)) Note too that what happens with clipping is device-dependent. If R's internal clipping is used, the part-circle is converted to a polygon. __ R-help@stat.math.ethz.ch 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. -- Dr Paul Murrell Department of Statistics The University of Auckland Private Bag 92019 Auckland New Zealand 64 9 3737599 x85392 [EMAIL PROTECTED] http://www.stat.auckland.ac.nz/~paul/ __ R-help@stat.math.ethz.ch 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] Percentage area of a distribution
It seems like this should be pretty straight forward, but for some reason the answer escapes me. I have a normal distribution S made up of two normal distributions C and C-bar. I need to find the percentage of the area of S that both C and C-bar occupy. Any suggestions? Thanks in advance, Alan Gibson __ R-help@stat.math.ethz.ch 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] Percentage area of a distribution
let me amend my previous message to remove a silly mistake: I have a (non-normal) distribution S that is a mixture of two normal distributions C and C-bar. I need to find the percentage of the area of S that both C and C-bar occupy. Thanks again, Alan Gibson On 5/1/07, Leeds, Mark (IED) [EMAIL PROTECTED] wrote: What do you mean by a normal distribution made up of to normal distributions ? A mixture of two Normals with different variances doesn't result in a normal distribution so I'm not sure what you Nean but maybe someone else does. No offense intended but, if noone replies, That means that noone else understood either. -Original Message- From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED] On Behalf Of Alan Gibson Sent: Tuesday, May 01, 2007 8:02 PM To: R-help@stat.math.ethz.ch Subject: [R] Percentage area of a distribution It seems like this should be pretty straight forward, but for some reason the answer escapes me. I have a normal distribution S made up of two normal distributions C and C-bar. I need to find the percentage of the area of S that both C and C-bar occupy. Any suggestions? Thanks in advance, Alan Gibson __ R-help@stat.math.ethz.ch 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. This is not an offer (or solicitation of an offer) to buy/sell the securities/instruments mentioned or an official confirmation. Morgan Stanley may deal as principal in or own or act as market maker for securities/instruments mentioned or may advise the issuers. This is not research and is not from MS Research but it may refer to a research analyst/research report. Unless indicated, these views are the author's and may differ from those of Morgan Stanley research or others in the Firm. We do not represent this is accurate or complete and we may not update this. Past performance is not indicative of future returns. For additional information, research reports and important disclosures, contact me or see https://secure.ms.com/servlet/cls. You should not use e-mail to request, authorize or effect the purchase or sale of any security or instrument, to send transfer instructions, or to effect any other transactions. We cannot guarantee that any such requests received vi! a e-mail will be processed in a timely manner. This communication is solely for the addressee(s) and may contain confidential information. We do not waive confidentiality by mistransmission. Contact me if you do not wish to receive these communications. In the UK, this communication is directed in the UK to those persons who are market counterparties or intermediate customers (as defined in the UK Financial Services Authority's rules). __ R-help@stat.math.ethz.ch 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] optimising fitted distributions
Dear R experts, I am relatively new to R and am interested in whether there is a package which will fit data with a swag of distributions and determine the optimal fit such as stat::fit from technical software solutions? As is discussed in the following r-help posting goodfit in vcd seems to do this for three of the distribution http://finzi.psych.upenn.edu/R/Rhelp02a/archive/37053.html, have there been any further developments for extending this to other distributions? Thanks in advanced Floris Floris van Ogtrop University of Sydney Dept. Food Agriculture and Natural Resources Australia [[alternative HTML version deleted]] __ R-help@stat.math.ethz.ch 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] ED50 from logistic model with interactions
Hi, I was wondering if someone could please help me. I am doing a logistic regression to compare size at maturity between 3 seasons. My model is: fit - glm(Mature ~ Season * Size - 1, family = binomial, data=dat) where Mature is a binary response, 0 for immature, 1 for mature. There are 3 Seasons. The Season * Size interaction is significant. I would like to compare the size at 50% maturity between Seasons, which I have calculated as: Mat50_S1 - -fit$coef[1]/fit$coef[4] Mat50_S2 - -fit$coef[2]/(fit$coef[4] + fit$coef[5]) Mat50_S3 - -fit$coef[3]/(fit$coef[4] + fit$coef[6]) But I am not sure how to calculate the standard error around each of these estimates. The p.dose function from the MASS package does this automatically, but it doesnt seem to allow interaction terms. In Faraway(2006) he has an example using the delta method to calculate the StdErr, but again without any interactions. I can apply this for the first Season, as there is just one intercept and one slope coefficient, but for the other 2 Seasons, the slope is a combination of the Size coefficient and the Size*Season coefficient, and I am not sure how to use the covariance matrix in the delta calculation. I could divide the data and do 3 different logistic regressions, one for each season, but while the Mat50 (i.e. mean Size at 50% maturity) is the same as that calculated by the separate lines regression, Im not sure how this may change the StdErr? Regards, Kate Kate Stark | PhD Candidate Institute of Antarctic Southern Ocean Studies Tasmanian Aquaculture Fisheries Institute University of Tasmania Email: kate.stark at utas.edu.au __ R-help@stat.math.ethz.ch 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] Concepts question: environment, frame, search path
Duncan: Thanks for taking a stab at my questions -- in following up I discovered the root of my difficulties -- I had not noticed document R-lang.pdf (R Language Definition). This clarifies a great deal. FWIW, it seems to me that a number of things I was hung up on (and which you discussed) revolved around: 1. Confusion between frame and data.frame. R-lang.pdf has several sections that touch on each of these, from which it's more clear (though not explicit) that these are not the same things. (Problematic: frame is mentioned first, is a more fundamental concept, yet has no entry in the Table of Contents, while data.frame does have an entry). (And the converse is true of the index!). 2. Ambiguity in the docs regarding environment, frame (and also regarding closely-related concepts closure and enclosure). Anyhow, I'm now in a much happier state :-). Regarding your questions: 1. data.frame: Ref[m]an p84: 'A data frame is a list of variables of the same length with unique row names, given class data.frame.' Which manual are you looking at? The reference index (refman.pdf)? [...] that statement may not be present in the current release Yes, the doc titled R: A Language and Environment for Statistical Computing Reference Index. This is in section I The base package, subsection data.frame, which was on page 84 of refman.pdf (which I downloaded yesterday, but now don't know where from) or on page 86 of fullrefman.pdf (downloaded today -- ie: current release). (And point understood on the suggestions about reporting doc issues -- though tracking them down to line numbers in the SVN is a bit optimistic, not to mention a moving target :-) --- Anyhow, thanks again for the response. Graham --- Graham Wideman Resources for programmable diagramming at: http://www.diagramantics.com Brain-related resources: http://wideman-one.com __ R-help@stat.math.ethz.ch 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] the Surv function
Hi, I'm trying to do a simple survival analysis on some data, and I'm having the following problem (here's my code and the error message): out - Surv(fup,event=status) Error in Surv(fup, event = status) : argument time2 is missing, with no default From reading the documentation, it seems that I should be able to simply write: Surv(time1, event) if my data is right-censored, which it is. Help! Thanks a million, Jen -- Jennifer Dillon Doctoral Student Harvard Biostatistics Room 414B, Building 1 [[alternative HTML version deleted]] __ R-help@stat.math.ethz.ch 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] the Surv function
Try using the following (without naming the arguments) - it should work. Surv(fup, status) I think what is happening is that if you use one or two arguments (unnamed), Surv assumes that you have right-censored data, with the first argument being time and the second being status. By specifying 'event' as you did, Surv assumes that you have 3 arguments of type=counting and it is looking for the ending time required by the counting data format and this is why you got the error. If you are not afraid of R code, type Surv (without the parentheses) to see the code behind the function. -Christos Christos Hatzis, Ph.D. Nuvera Biosciences, Inc. 400 West Cummings Park Suite 5350 Woburn, MA 01801 Tel: 781-938-3830 www.nuverabio.com -Original Message- From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED] On Behalf Of Jennifer Dillon Sent: Tuesday, May 01, 2007 11:16 PM To: r-help@stat.math.ethz.ch Subject: [R] the Surv function Hi, I'm trying to do a simple survival analysis on some data, and I'm having the following problem (here's my code and the error message): out - Surv(fup,event=status) Error in Surv(fup, event = status) : argument time2 is missing, with no default From reading the documentation, it seems that I should be able to simply write: Surv(time1, event) if my data is right-censored, which it is. Help! Thanks a million, Jen -- Jennifer Dillon Doctoral Student Harvard Biostatistics Room 414B, Building 1 [[alternative HTML version deleted]] __ R-help@stat.math.ethz.ch 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-help@stat.math.ethz.ch 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] ? R 2.5.0 alpha bug
This email is intended to highlight 2 problems that I encountered running R 2.5.0 alpha on a Windows XP machine. #1 - Open script error If I click the Open folder icon on the toolbar, R opens my script files perfectly. However, when I select File Open Script MyFileLocation, I get a fatal error that causes R to close immediately. This error was reproduced on 3 consecutive occasions but has been intermittent thereafter. One of these fatal errors resulted in a typical error reporting box being generated which I sent off. I was not able to verify if this error has been reported and corrected in subsequent versions of 2.5. #2 - Bug reporting link on CRAN website broken I tried to report the bug listed above on the CRAN website but when I clicked on the bug reporting link on the left-hand side panel of the main site (http://bugs.r-project.org/cgi-bin/R) , I get an error page with the following message: The system encountered a fatal error cannot open config file /home/sfe/r-bugs/jitterbug/R : No such file or directory The last error code was: No such file or directory uid/gid=30/8 This has been submitted to r-devel. Brant Inman __ R-help@stat.math.ethz.ch 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] Warnings in package dependencies and /src contains object files.
Dear All, I recently wrote a package in R and did a check on my package, all is well except for two warnings: * checking package dependencies ... WARNING 'library' or 'require' calls not declared from: MASS See the information on DESCRIPTION files in the chapter 'Creating R packages' of the 'Writing R Extensions' manual. * checking if this is a source package ... WARNING Subdirectory 'GLDEX/src' contains object files. In my package I have only used the dataset from MASS by way of examples, I wonder if it is customary to include all the libraries even if you only use them as examples? The second warning I am not sure about, I wonder if it just means I have more files other than the source codes in the /src folder? Thanks in advance for your comments. Steve. [[alternative HTML version deleted]] __ R-help@stat.math.ethz.ch 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.