Re: [R] Making R packages (Unix)
Gattuso, Jean-Pierre wrote: Hi: I have have taken over from a colleague who prepared an R package and failed to build it on Windows. I am doing this with unix as I am a mac user. Below is the output I get when I use the build command: [gattuso:unix/R/CO2.Rcheck] gattuso% R CMD build CO2 * checking for file 'CO2/DESCRIPTION' ... OK * preparing 'CO2': * checking whether 'INDEX' is up-to-date ... NO * use '--force' to overwrite the existing 'INDEX' * removing junk files * building 'CO2_1.0.tar.gz' tar: Unable to access /Users/gattuso/documents/unix/R/CO2.Rcheck/CO2_1.0.tar No such file or directory tar: WARNING! These file names were not selected: /Users/gattuso/documents/unix/R/CO2.Rcheck/CO2_1.0.tar CO2_1.0.tar: No such file or directory Looks like R_HOME is set to an invalid directory. Regards Michael -- Michael T. Mader Institute for Bioinformatics/MIPS, GSF Ingolstaedter Landstrasse 1 D-80937 Neuherberg 0049-89-3187-3576 In statistics, some people worry about not seeing the forest for the trees. I like to look at the bark. (C. R. Blyth, 1967) __ [EMAIL PROTECTED] mailing list https://www.stat.math.ethz.ch/mailman/listinfo/r-help
[R] Need your help-SOS
Hello, I am a newbie in R project and trying to call prcomp(x) of R function using (D)COM server communicate with R in ASP, and encountering the error Runtime error -2147221493(8004000b). Automation Error, Object is static, operation not allowed. Source code is shown as below: % Set StatConn=Server.CreateObject(StatConnectorSrv.StatConnector) StatConn.Init (R) Result=StatConn.Evaluate(x-matrix(c(1,2,3,4,5,6,7,8,9),3)) Result=StatConn.Evaluate(y-prcomp(x)) StatConn.Close % I have another problem when displaying dimension variable Result(1,1) on client, there will be a Type mismatch: 'Result' error. Source code is shown as below: % Set StatConn=Server.CreateObject(StatConnectorSrv.StatConnector) StatConn.Init (R) Result=StatConn.Evaluate(x-matrix(c(1,2,3,4,5,6,7,8,9),3)) Response.write Result(1,1) StatConn.Close % You are very appreciated if you help me on above issue. Thanks Again! Regards, WeiQiang Li IT-Factory Information Systems Tel: 6485-2439 __ [EMAIL PROTECTED] mailing list https://www.stat.math.ethz.ch/mailman/listinfo/r-help
[R] C code for KalmnaLike
Hi it is possible to see the C code for the KalmanLike and Kalmansmooth functions with R? Otherwise, without using R, how can I get the code? Thank arianna __ [EMAIL PROTECTED] mailing list https://www.stat.math.ethz.ch/mailman/listinfo/r-help
[R] geoR variogram problem
Dear GeoR-er, If I use the variog function in the latest release of geoR, the first lag is always ignored. For instance, if you read in geodata, calculates the variogram using the variog function and give in a uvec like uvec=seq(0,max,by=2.44), it only starts giving results from distance=4.88 and ignores 2.44! This wasn't the case in former versions of geoR. Is this done on purpose? Why? Tnx for your help, Dave Nys K.U.Leuven Laboratory for Forest, Nature and Landscape Research Vital De Costerstraat 102 B-3000 Leuven Tel. +32 (0)16 329751 Fax. +32 (0)16 329760 email: [EMAIL PROTECTED] __ [EMAIL PROTECTED] mailing list https://www.stat.math.ethz.ch/mailman/listinfo/r-help
[R] Need your help with SJava package on W2K
Dear R expert, I have the problems with running R from Java on Windows 2000. This is my what i get when i run the program: Loading RInterpreter library java.lang.UnsatisfiedLinkError: no RInterpreter in java.library.path I set all variable (environment). I'd like to notice,that Java from R works fine, initialization was without any problems. Please help me to solve me this problem. Raimondas __ [EMAIL PROTECTED] mailing list https://www.stat.math.ethz.ch/mailman/listinfo/r-help
Re: [R] Need your help-SOS
WeiQiang - As I read it, both difficulties arise on the DCOM side, not in the R syntax. Problem 1, and I'm just guessing, could arise if you are not allowed to overwrite the value of Result in the DCOM environment. Try again, using two different variable names in the two successive lines. Problem 2 would seem to have something to do with the DCOM syntax for subscripting an array. Inside R, the syntax would be to use square brackets for subscripting x, thus, x[1,1] would return the upper left corner element. I have no knowledge of DCOM, so I don't know how the command Response.write Result(1,1) gets interpreted. - tom blackwell - u michigan medical school - ann arbor - On Wed, 10 Sep 2003 [EMAIL PROTECTED] wrote: Hello, I am a newbie in R project and trying to call prcomp(x) of R function using (D)COM server communicate with R in ASP, and encountering the error Runtime error -2147221493(8004000b). Automation Error, Object is static, operation not allowed. Source code is shown as below: % Set StatConn=Server.CreateObject(StatConnectorSrv.StatConnector) StatConn.Init (R) Result=StatConn.Evaluate(x-matrix(c(1,2,3,4,5,6,7,8,9),3)) Result=StatConn.Evaluate(y-prcomp(x)) StatConn.Close % I have another problem when displaying dimension variable Result(1,1) on client, there will be a Type mismatch: 'Result' error. Source code is shown as below: % Set StatConn=Server.CreateObject(StatConnectorSrv.StatConnector) StatConn.Init (R) Result=StatConn.Evaluate(x-matrix(c(1,2,3,4,5,6,7,8,9),3)) Response.write Result(1,1) StatConn.Close % You are very appreciated if you help me on above issue. Thanks Again! Regards, WeiQiang Li IT-Factory Information Systems Tel: 6485-2439 __ [EMAIL PROTECTED] mailing list https://www.stat.math.ethz.ch/mailman/listinfo/r-help __ [EMAIL PROTECTED] mailing list https://www.stat.math.ethz.ch/mailman/listinfo/r-help
[R] Off Topic: Good reference for sample size calculations
Hi All, This is off topic, but we're drawing a blank here.. In a presentation I'll be giving next week, I want to include a reference to a good general text on computing sample sizes for standard experiments. Can anyone recommend a good book to use for this purpose? Thanks, -Greg LEGAL NOTICE\ Unless expressly stated otherwise, this messag...{{dropped}} __ [EMAIL PROTECTED] mailing list https://www.stat.math.ethz.ch/mailman/listinfo/r-help
RE: [R] Off Topic: Good reference for sample size calculations
Jacob Cohen's book Statistical Power Analysis for the Behavioral Sciences is one. -- Harold C. Doran Director of Research and Evaluation New American Schools 675 N. Washington Street, Suite 220 Alexandria, Virginia 22314 703.647.1628 http://www.edperform.net -Original Message- From: Warnes, Gregory R [mailto:[EMAIL PROTECTED] Sent: Wednesday, September 10, 2003 8:37 AM To: '[EMAIL PROTECTED]' Subject: [R] Off Topic: Good reference for sample size calculations Hi All, This is off topic, but we're drawing a blank here.. In a presentation I'll be giving next week, I want to include a reference to a good general text on computing sample sizes for standard experiments. Can anyone recommend a good book to use for this purpose? Thanks, -Greg LEGAL NOTICE\ Unless expressly stated otherwise, this messag...{{dropped}} __ [EMAIL PROTECTED] mailing list https://www.stat.math.ethz.ch/mailman/listinfo/r-help __ [EMAIL PROTECTED] mailing list https://www.stat.math.ethz.ch/mailman/listinfo/r-help
Re: [R] Off Topic: Good reference for sample size calculations
On Wed, 2003-09-10 at 07:37, Warnes, Gregory R wrote: Hi All, This is off topic, but we're drawing a blank here.. In a presentation I'll be giving next week, I want to include a reference to a good general text on computing sample sizes for standard experiments. Can anyone recommend a good book to use for this purpose? Thanks, -Greg Greg, How about: Design and Analysis of Clinical Trials Concepts and Methodologies Shein-Chung Chow and Jen-pei Liu Wiley, 1998 Chapter 10: Sample Size Determination Pages 424 - 482 Amazon.com link: http://www.amazon.com/exec/obidos/tg/detail/-/047113404X/ HTH, Marc Schwartz __ [EMAIL PROTECTED] mailing list https://www.stat.math.ethz.ch/mailman/listinfo/r-help
Re: [R] C code for KalmnaLike
Did you try www.r-project.org - Download CRAN - {select a local mirror} - Source code for all platforms? From what I hear, your prayers should be answered there (though I have no personnally built links to compiled code since S-Plus 3.3). hope this helps. spencer graves [EMAIL PROTECTED] wrote: Hi it is possible to see the C code for the KalmanLike and Kalmansmooth functions with R? Otherwise, without using R, how can I get the code? Thank arianna __ [EMAIL PROTECTED] mailing list https://www.stat.math.ethz.ch/mailman/listinfo/r-help __ [EMAIL PROTECTED] mailing list https://www.stat.math.ethz.ch/mailman/listinfo/r-help
Re: [R] C code for KalmnaLike
[EMAIL PROTECTED] wrote: Hi it is possible to see the C code for the KalmanLike and Kalmansmooth functions with R? It is possible to see all the code distributed with R and the R packages on CRAN. (This is why it is sometimes called an open source project.) Kalman smoothing and Kalman filtering, on which the likelihood calculation is usually based, are not part of base R, but are in packages. I am not sure about other packages, but the versions in the dse package are not written in C. There is a fortran version and an equivalent R version. There is a C version which is translated with f2c, but that is not the place to look if you are trying to understand the algorithm. The easiest code to read is the R version, but the fortran it the one that is used (by default) for reason of speed. Otherwise, without using R, how can I get the code? You can download the source from CRAN and examine it without using R. However, if you actually want to use the code, then I strongly suggest you use R too. The R code makes a call to fortran in order to speed the iterative part of the calculation, but the likelihood calculation from the residuals, all the error checking, plotting, and nice to use features are in the R code. If you want to use dse, then start by reading the dse user's guide distributed with the package bundle (in dse1/inst/doc/dse-guide.pdf) and getting familiar with R. (When I have been asked questions like your's before, the next question has usually be something like: please give me a tutorial on the internal details of your code, because I want to pull it apart and use it somewhere else. I am not very interested in the effort required for me to do that.) Paul Gilbert Thank arianna __ [EMAIL PROTECTED] mailing list https://www.stat.math.ethz.ch/mailman/listinfo/r-help
[R] Plot survey data
I am trying to make plots that take into account survey weights. This a survey of the US population. To start with I want to explore the data using pairs, plot, coplots and lattice. Are there specialized methods that handle survey weights for plotting? Any pointers? Anupam. [[alternative HTML version deleted]] __ [EMAIL PROTECTED] mailing list https://www.stat.math.ethz.ch/mailman/listinfo/r-help
Re: [R] geoR variogram problem
Dave GHard to tell withoiut see your data. Pleas send me and example code (and data if necessary) and I will chack P.J. On Wed, 10 Sep 2003, Dave Nys wrote: Dear GeoR-er, If I use the variog function in the latest release of geoR, the first lag is always ignored. For instance, if you read in geodata, calculates the variogram using the variog function and give in a uvec like uvec=seq(0,max,by=2.44), it only starts giving results from distance=4.88 and ignores 2.44! This wasn't the case in former versions of geoR. Is this done on purpose? Why? Tnx for your help, Dave Nys K.U.Leuven Laboratory for Forest, Nature and Landscape Research Vital De Costerstraat 102 B-3000 Leuven Tel. +32 (0)16 329751 Fax. +32 (0)16 329760 email: [EMAIL PROTECTED] __ [EMAIL PROTECTED] mailing list https://www.stat.math.ethz.ch/mailman/listinfo/r-help Paulo Justiniano Ribeiro Jr Departamento de EstatÃstica Universidade Federal do Paraná Caixa Postal 19.081 CEP 81.531-990 Curitiba, PR - Brasil Tel: (+55) 41 361 3471 Fax: (+55) 41 361 3141 e-mail: [EMAIL PROTECTED] http://www.est.ufpr.br/~paulojus __ [EMAIL PROTECTED] mailing list https://www.stat.math.ethz.ch/mailman/listinfo/r-help
[R] sort a matrix on just one column
How can I sort(decreasing) a matrix on just the first column? For example, I can I get 8 2 7 5 4 1 from 7 5 4 1 8 2 Thanks __ [EMAIL PROTECTED] mailing list https://www.stat.math.ethz.ch/mailman/listinfo/r-help
Re: [R] Making R packages (Unix)
On Tue, 9 Sep 2003, Gattuso, Jean-Pierre wrote: Hi: I have have taken over from a colleague who prepared an R package and failed to build it on Windows. I am doing this with unix as I am a mac user. Below is the output I get when I use the build command: This looks like what happens under Mac OS X, where tar doesn't recognise the -X flag for eXcluding files [it's used for something else]. You can confirm by checking man tar. I don't know if this will get fixed for 1.8.0, but a work-around is to get GNU tar. -thomas [gattuso:unix/R/CO2.Rcheck] gattuso% R CMD build CO2 * checking for file 'CO2/DESCRIPTION' ... OK * preparing 'CO2': * checking whether 'INDEX' is up-to-date ... NO * use '--force' to overwrite the existing 'INDEX' * removing junk files * building 'CO2_1.0.tar.gz' tar: Unable to access /Users/gattuso/documents/unix/R/CO2.Rcheck/CO2_1.0.tar No such file or directory tar: WARNING! These file names were not selected: /Users/gattuso/documents/unix/R/CO2.Rcheck/CO2_1.0.tar CO2_1.0.tar: No such file or directory I have read the R-exts.pdf document but did not find what I do wrong. In fact, there is little information on the use of R CMD build, both in the R-exts.pdf document and the man (R CMD build --help) pages. Do I need to make the CO2_1.0.tar file myself? Your help would be much appreciated! jp -- Jean-Pierre Gattuso | mailto:[EMAIL PROTECTED] | http://www.obs-vlfr.fr/~gattuso __ [EMAIL PROTECTED] mailing list https://www.stat.math.ethz.ch/mailman/listinfo/r-help Thomas Lumley Assoc. Professor, Biostatistics [EMAIL PROTECTED] University of Washington, Seattle __ [EMAIL PROTECTED] mailing list https://www.stat.math.ethz.ch/mailman/listinfo/r-help
Re: [R] sort a matrix on just one column
Hi On 10 Sep 2003 at 10:50, Paul Green wrote: How can I sort(decreasing) a matrix on just the first column? For example, I can I get 8 2 7 5 4 1 from 7 5 4 1 8 2 I am sure in help pages for sort() is a link to order() mat x y [1,] 7 5 [2,] 4 1 [3,] 8 2 o-order(mat[,1],decreasing=T) mat[o,] x y [1,] 8 2 [2,] 7 5 [3,] 4 1 Thanks __ [EMAIL PROTECTED] mailing list https://www.stat.math.ethz.ch/mailman/listinfo/r-help CheersPetr Pikal [EMAIL PROTECTED] __ [EMAIL PROTECTED] mailing list https://www.stat.math.ethz.ch/mailman/listinfo/r-help
[R] dataframe subsetting
I can create a small dataset, x below, and subset out rows based on values of a certain variable. However, on the dataset I'm working on now, latdata below, I get a subscript error. Any advice is appreciated! Ryan Successful: is.data.frame(x) [1] TRUE x X1 X2 X3 1 1 3 5 2 2 4 6 x[x$X2 %in% c(3),] X1 X2 X3 1 1 3 5 Unsuccessful: is.data.frame(latdata) [1] TRUE is.numeric(latdata$intent) [1] TRUE table(latdata$intent) 1 2 3 4 5 6 34 23 67 179 996 2 unlikely - latdata[latdata$intent %in% c(1,2,3,4),] Error in x[[j]] : subscript out of bounds -- Ryan T. Moore ~ Government Social Policy Ph.D. Candidate ~ Harvard University __ [EMAIL PROTECTED] mailing list https://www.stat.math.ethz.ch/mailman/listinfo/r-help
Re: [R] Plot survey data
Dear Anupam, I may be wrong, but I don't think that there's any standard method to use in plotting with case weights. I can think of two approaches, however: (1) If you have a large sample, and if the range of the weights isn't too large, you could sample your observations with probability of inclusion in the plot proportional to the case weights. (2) You could plot the points with size proportional to the square root of the case weights (i.e., area proportional to the weights). I hope that this helps, John At 10:46 AM 9/10/2003 -0400, [EMAIL PROTECTED] wrote: I am trying to make plots that take into account survey weights. This a survey of the US population. To start with I want to explore the data using pairs, plot, coplots and lattice. Are there specialized methods that handle survey weights for plotting? Any pointers? Anupam. - John Fox Department of Sociology McMaster University Hamilton, Ontario, Canada L8S 4M4 email: [EMAIL PROTECTED] phone: 905-525-9140x23604 web: www.socsci.mcmaster.ca/jfox __ [EMAIL PROTECTED] mailing list https://www.stat.math.ethz.ch/mailman/listinfo/r-help
[R] what is set.fit in function predict.lm
Hi, what es the parameter set.fit in function predict.lm, is set.fit True then i need the standard error How i cant calculate it?. It is the different what? i see the code of predict.lm How i cant see the matemathics formula for the calculation of standard error. __ [EMAIL PROTECTED] mailing list https://www.stat.math.ethz.ch/mailman/listinfo/r-help
Re: [R] C code for KalmnaLike
My apologies. I now realize this question was more specifically about the KalmanLike and KalmanSmooth functions distributed in package ts with base, and not generally about code for calculating the Kalman smoother or likelihood from the Kalman filter. Of course, the sources for these functions are available and appear to be written in C. Paul Gilbert Paul Gilbert wrote: [EMAIL PROTECTED] wrote: Hi it is possible to see the C code for the KalmanLike and Kalmansmooth functions with R? It is possible to see all the code distributed with R and the R packages on CRAN. (This is why it is sometimes called an open source project.) Kalman smoothing and Kalman filtering, on which the likelihood calculation is usually based, are not part of base R, but are in packages. I am not sure about other packages, but the versions in the dse package are not written in C. There is a fortran version and an equivalent R version. There is a C version which is translated with f2c, but that is not the place to look if you are trying to understand the algorithm. The easiest code to read is the R version, but the fortran it the one that is used (by default) for reason of speed. Otherwise, without using R, how can I get the code? You can download the source from CRAN and examine it without using R. However, if you actually want to use the code, then I strongly suggest you use R too. The R code makes a call to fortran in order to speed the iterative part of the calculation, but the likelihood calculation from the residuals, all the error checking, plotting, and nice to use features are in the R code. If you want to use dse, then start by reading the dse user's guide distributed with the package bundle (in dse1/inst/doc/dse-guide.pdf) and getting familiar with R. (When I have been asked questions like your's before, the next question has usually be something like: please give me a tutorial on the internal details of your code, because I want to pull it apart and use it somewhere else. I am not very interested in the effort required for me to do that.) Paul Gilbert Thank arianna __ [EMAIL PROTECTED] mailing list https://www.stat.math.ethz.ch/mailman/listinfo/r-help __ [EMAIL PROTECTED] mailing list https://www.stat.math.ethz.ch/mailman/listinfo/r-help
Re: [R] insert eps into microsft word
If you're working in windows, the WMF (windows metafile) format is probably your best bet; it's a vector format like PostScript. On Wed, 10 Sep 2003, Karim Elsawy wrote: it seems that word can not read encapsupalted postscripts generated by R I used this command postscript(output.eps,horizontal=F,onefile=TRUE) since onefile=TRUE produces an encapsualted postscript actually what I'm trying to do is to insert the postsript file into a word document since other formats like jpeg and bmp do not reproduce the same quality like postscript formats any suggestions are very much appreciated Karim __ [EMAIL PROTECTED] mailing list https://www.stat.math.ethz.ch/mailman/listinfo/r-help -- 620B Bartram Hall[EMAIL PROTECTED] Zoology Department, University of Floridahttp://www.zoo.ufl.edu/bolker Box 118525 (ph) 352-392-5697 Gainesville, FL 32611-8525 (fax) 352-392-3704 __ [EMAIL PROTECTED] mailing list https://www.stat.math.ethz.ch/mailman/listinfo/r-help
Re: [R] insert eps into microsft word
On Wed, 2003-09-10 at 12:12, Karim Elsawy wrote: it seems that word can not read encapsupalted postscripts generated by R I used this command postscript(output.eps,horizontal=F,onefile=TRUE) since onefile=TRUE produces an encapsualted postscript actually what I'm trying to do is to insert the postsript file into a word document since other formats like jpeg and bmp do not reproduce the same quality like postscript formats any suggestions are very much appreciated Karim Try this syntax: postscript(output.eps, horizontal = FALSE, onefile = FALSE, paper = special) Take note of the instructions in the Details section of ?postscript and also re-read the description of 'onefile' ('**' are my add): logical: if true (the default) allow multiple figures in one file. **If false**, generate a file name containing the page number and use an EPSF header and no DocumentMedia comment. HTH, Marc Schwartz __ [EMAIL PROTECTED] mailing list https://www.stat.math.ethz.ch/mailman/listinfo/r-help
[R] scan() problem
Dear R-helpers, I have to read some large csv-files into R (30 - 100MB). Since reading with read.csv leads to memory exhausted, I tried with scan(), skipping not needed columns by NULL-elements in what. When these skipped elements are quoted strings with commata inside, R interprets each such quoted comma as element separator leading to wrong records in the rest of the line. A little test will show what I mean. I have the following test.csv: col.A,col.B,col.C,col.D 1,quoted string,again, again again,123 2,nice quotes, isnt it,you got it,456 First I read all elements: tst - scan(test.csv, what=list(a=0,b=,c=,d=0), sep=,, skip=1) Read 2 records tst $a [1] 1 2 $b [1] quoted stringnice quotes, isnt it $c [1] again, again again you got it $d [1] 123 456 Everything is fine. Then I try to skip the 2nd column by giving b=NULL: tst - scan(test.csv, what=list(a=0,b=NULL,c=,d=0), sep=,, skip=1) Read 2 records Warning message: number of items read is not a multiple of the number of columns tst $a [1] 1 2 $b NULL $c [1] again, again again isnt it,you got it,456\n\n\n $d [1] 123 NA I got garbage. Isn't this a bug? Or did I something wrong? Is there a workaround? Thank you all, Paul Bayer, Feldafing, Germany __ [EMAIL PROTECTED] mailing list https://www.stat.math.ethz.ch/mailman/listinfo/r-help
[R] PLS LDA
Dear R experts I saw and downloaded the fresh pls package for R. Is there any way of using this pls package for PLS discriminant analysis? If not, is there any other package available. I need a way of classifying objects into e.g. two groups, where nbr_observations nbr_variables many thanks for your kind help Christoph -- Christoph Lehmann [EMAIL PROTECTED] __ [EMAIL PROTECTED] mailing list https://www.stat.math.ethz.ch/mailman/listinfo/r-help
Re: [R] insert eps into microsft word
You can also add a Windows metafile preview to your eps image using for instance Ghostscript. Martin On Wed, 10 Sep 2003 18:12:00 +0100, Karim Elsawy [EMAIL PROTECTED] wrote: it seems that word can not read encapsupalted postscripts generated by R I used this command postscript(output.eps,horizontal=F,onefile=TRUE) since onefile=TRUE produces an encapsualted postscript actually what I'm trying to do is to insert the postsript file into a word document since other formats like jpeg and bmp do not reproduce the same quality like postscript formats any suggestions are very much appreciated Karim __ [EMAIL PROTECTED] mailing list https://www.stat.math.ethz.ch/mailman/listinfo/r-help -- Martin Biuw Sea Mammal Research Unit Gatty Marine Laboratory, University of St Andrews St Andrews, Fife KY16 8PA Scotland Ph: +44-(0)1334-462637 Fax: +44-(0)1334-462632 Web: http://smub.st.and.ac.uk __ [EMAIL PROTECTED] mailing list https://www.stat.math.ethz.ch/mailman/listinfo/r-help
[R] logistic regression for a data set with perfect separation
Dear R experts I have the follwoing data V1 V2 1 -5.800 0 2 -4.800 0 3 -2.867 0 4 -0.867 0 5 -0.733 0 6 -1.667 0 7 -0.133 1 8 1.200 1 9 1.333 1 and I want to know, whether V1 can predict V2: of course it can, since there is a perfect separation between cases 1..6 and 7..9 How can I test, whether this conclusion (being able to assign an observation i to class j, only knowing its value on Variable V1) holds also for the population, our data were drawn from? Means, which inference procedure is recommended? Logistic regression, for obvious reasons makes no sense. Many thanks for your help Christoph -- Christoph Lehmann [EMAIL PROTECTED] __ [EMAIL PROTECTED] mailing list https://www.stat.math.ethz.ch/mailman/listinfo/r-help
Re: [R] logistic regression for a data set with perfect separation
On Wednesday, Sep 10, 2003, at 18:50 Europe/London, Christoph Lehmann wrote: Dear R experts I have the follwoing data V1 V2 1 -5.800 0 2 -4.800 0 3 -2.867 0 4 -0.867 0 5 -0.733 0 6 -1.667 0 7 -0.133 1 8 1.200 1 9 1.333 1 and I want to know, whether V1 can predict V2: of course it can, since there is a perfect separation between cases 1..6 and 7..9 How can I test, whether this conclusion (being able to assign an observation i to class j, only knowing its value on Variable V1) holds also for the population, our data were drawn from? For this you really need more data. The only way you'll ever be able to reject that hypothesis is by finding an instance of 010 or 101 in the (ordered by V1) sample. And if you find such then you can reject with certainty. Means, which inference procedure is recommended? Logistic regression, for obvious reasons makes no sense. Not so obvious to me! Logistic regression still makes sense, but care is needed in the method of estimation/inference. The maximum likelihood solution in the above case is a model which says V2 is 1 with certainty at some values of V1, and is zero with certainty at other values; and that seems an unwarranted inference with so little data. That's a criticism of maximum likelihood, rather than a criticism of logistic regression. (Think about the more extreme situation of tossing a coin once: if a head is observed, the ML solution is that the coin lands heads with certainty, ie that there no chance of tails.) There are alternative (Bayesian and pseudo-Bayesian) methods of inference which can yield more sensible answers in general. [One such is implemented in package brlr (bias reduced logistic regression) on CRAN.] To test the hypothesis described above, though, with the data you have, would seem to require a fully Bayesian analysis whose conclusions would depend strongly on the prior probability attached to the hypothesis. ie you need more data... I hope that helps in some way! Regards, David Many thanks for your help Christoph -- Christoph Lehmann [EMAIL PROTECTED] __ [EMAIL PROTECTED] mailing list https://www.stat.math.ethz.ch/mailman/listinfo/r-help __ [EMAIL PROTECTED] mailing list https://www.stat.math.ethz.ch/mailman/listinfo/r-help
[R] how to calculate Spearman correlation with missing values
Hello, there: I got data matix with missing values. I want to calculate any possible pairwise Spearman correlation rho for each column. Is there a function just like cor(x, y, use=complete.obs) for Pearson correlation? Thanks in advance! Josh __ [EMAIL PROTECTED] mailing list https://www.stat.math.ethz.ch/mailman/listinfo/r-help
Re: [R] how to calculate Spearman correlation with missing values
On Wed, 10 Sep 2003 [EMAIL PROTECTED] wrote: Hello, there: I got data matix with missing values. I want to calculate any possible pairwise Spearman correlation rho for each column. Is there a function just like cor(x, y, use=complete.obs) for Pearson correlation? Thanks in advance! ?cor.test Josh __ [EMAIL PROTECTED] mailing list https://www.stat.math.ethz.ch/mailman/listinfo/r-help __ [EMAIL PROTECTED] mailing list https://www.stat.math.ethz.ch/mailman/listinfo/r-help
Re: [R] insert eps into microsft word
Depending on your version of Word that simply is *not* true! I can't test this here (I now have Word XP/2002) but my colleagues have Word 97 on their college-supplied service (yes, really!) and despite not being able to see a preview, they can quite happily import eps files I have produced during the course of my work. Admitedly a blank box in word is not the most user-friendly thing to work with, but this old version of Word will happily deal with the postscript produced. Having used Word 2000 and XP during the course of writing a PhD thesis that used R-generated plots extensively, I know first hand how easy it is to deal with eps files from R in Word. My setup involves using Ghostscript/GSView on Windows XP to view the eps plots I produce in R and then import them into Word XP (the eps filter now seems to add a preview for you), but if you have an earlier word version, GSView can easily add a preview for you. Also, onefile = TRUE does *not* produce eps files. I quote from ?postscript : onefile: logical: if true (the default) allow multiple figures in one file. If false, generate a file name containing the page number and use an EPSF header and no `DocumentMedia' comment. It is a bit confusing, but onefile = TRUE is for producing multi-page postscript documents, i.e. put all the plots to follow into a single postscript doc with (possibly) more than a single page. I know a number of people have suggested using a windows metafile. Whilst this might offer a solution to your problem, unless you are using a very-out-of-date version of Word I cannot see the advantages of using metafiles over postscript. HTH Gavin Karim Elsawy wrote: it seems that word can not read encapsupalted postscripts generated by R I used this command postscript(output.eps,horizontal=F,onefile=TRUE) since onefile=TRUE produces an encapsualted postscript actually what I'm trying to do is to insert the postsript file into a word document since other formats like jpeg and bmp do not reproduce the same quality like postscript formats any suggestions are very much appreciated Karim __ [EMAIL PROTECTED] mailing list https://www.stat.math.ethz.ch/mailman/listinfo/r-help -- %~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~% Gavin Simpson [T] +44 (0)20 7679 5522 ENSIS Research Fellow [F] +44 (0)20 7679 7565 ENSIS Ltd. ECRC [E] [EMAIL PROTECTED] UCL Department of Geography [W] http://www.ucl.ac.uk/~ucfagls/cv/ 26 Bedford Way[W] http://www.ucl.ac.uk/~ucfagls/ London. WC1H 0AP. %~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~% __ [EMAIL PROTECTED] mailing list https://www.stat.math.ethz.ch/mailman/listinfo/r-help
Re: [R] insert eps into microsft word
Rafael A. Irizarry [EMAIL PROTECTED] writes: for word documents submitted to picky journals i usually use seomthing like this: bitmap(plot_1.png,width=6,height=6,res=600,pointsize=12,family=Times) on my computer this resutls in quality just as good (to mu eye) as with postscript. you can also use adobe acrobat to convert postsctipt to something else. Word seems generally unhappy with PostScript, not just the files R makes. One option that I have seen come out rather nicely is to have the plots as PDF and use Adobe Distiller as the backend so that the whole thing becomes a PDF file. -- O__ Peter Dalgaard Blegdamsvej 3 c/ /'_ --- Dept. of Biostatistics 2200 Cph. N (*) \(*) -- University of Copenhagen Denmark Ph: (+45) 35327918 ~~ - ([EMAIL PROTECTED]) FAX: (+45) 35327907 __ [EMAIL PROTECTED] mailing list https://www.stat.math.ethz.ch/mailman/listinfo/r-help
Re: [R] insert eps into microsft word
Your problem may also be that you can't get the figure to print from Word? To get Word to print anything other than a blank box for the eps, you will need to install and use a postscript printer driver for your printer. cheers Brett Brett Melbourne, Postdoctoral Fellow Biological Invasions IGERT www.cpb.ucdavis.edu/bioinv Center for Population Biology, Storer Hall University of California Davis CA 95616 - Original Message - From: Gavin Simpson [EMAIL PROTECTED] To: Karim Elsawy [EMAIL PROTECTED] Cc: [EMAIL PROTECTED] Sent: Wednesday, September 10, 2003 12:26 PM Subject: Re: [R] insert eps into microsft word Depending on your version of Word that simply is *not* true! I can't test this here (I now have Word XP/2002) but my colleagues have Word 97 on their college-supplied service (yes, really!) and despite not being able to see a preview, they can quite happily import eps files I have produced during the course of my work. Admitedly a blank box in word is not the most user-friendly thing to work with, but this old version of Word will happily deal with the postscript produced. Having used Word 2000 and XP during the course of writing a PhD thesis that used R-generated plots extensively, I know first hand how easy it is to deal with eps files from R in Word. My setup involves using Ghostscript/GSView on Windows XP to view the eps plots I produce in R and then import them into Word XP (the eps filter now seems to add a preview for you), but if you have an earlier word version, GSView can easily add a preview for you. Also, onefile = TRUE does *not* produce eps files. I quote from ?postscript : onefile: logical: if true (the default) allow multiple figures in one file. If false, generate a file name containing the page number and use an EPSF header and no `DocumentMedia' comment. It is a bit confusing, but onefile = TRUE is for producing multi-page postscript documents, i.e. put all the plots to follow into a single postscript doc with (possibly) more than a single page. I know a number of people have suggested using a windows metafile. Whilst this might offer a solution to your problem, unless you are using a very-out-of-date version of Word I cannot see the advantages of using metafiles over postscript. HTH Gavin Karim Elsawy wrote: it seems that word can not read encapsupalted postscripts generated by R I used this command postscript(output.eps,horizontal=F,onefile=TRUE) since onefile=TRUE produces an encapsualted postscript actually what I'm trying to do is to insert the postsript file into a word document since other formats like jpeg and bmp do not reproduce the same quality like postscript formats any suggestions are very much appreciated Karim __ [EMAIL PROTECTED] mailing list https://www.stat.math.ethz.ch/mailman/listinfo/r-help -- %~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~% Gavin Simpson [T] +44 (0)20 7679 5522 ENSIS Research Fellow [F] +44 (0)20 7679 7565 ENSIS Ltd. ECRC [E] [EMAIL PROTECTED] UCL Department of Geography [W] http://www.ucl.ac.uk/~ucfagls/cv/ 26 Bedford Way[W] http://www.ucl.ac.uk/~ucfagls/ London. WC1H 0AP. %~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~% __ [EMAIL PROTECTED] mailing list https://www.stat.math.ethz.ch/mailman/listinfo/r-help __ [EMAIL PROTECTED] mailing list https://www.stat.math.ethz.ch/mailman/listinfo/r-help
RE: [R] logistic regression for a data set with perfect separati
On 10-Sep-03 Christoph Lehmann wrote: I have the follwoing data V1 V2 1 -5.800 0 2 -4.800 0 3 -2.867 0 4 -0.867 0 5 -0.733 0 6 -1.667 0 7 -0.133 1 8 1.200 1 9 1.333 1 and I want to know, whether V1 can predict V2: of course it can, since there is a perfect separation between cases 1..6 and 7..9 How can I test, whether this conclusion (being able to assign an observation i to class j, only knowing its value on Variable V1) holds also for the population, our data were drawn from? Means, which inference procedure is recommended? Logistic regression, for obvious reasons makes no sense. This is not so much an R question, nor really a which procedure question, since standard procedures are not usually equipped to deal with such situations (beyond telling you in some way that the situation is out of bounds). However, you can certainly investigate it by writing little R programs to look at it from various points of view. Let 'm' denote the location parameter for the CDF which models the probability, and 's' the scale parameter (e.g. a logistic function). For a start, clearly the maximum of the likelihood is 1, achieved when s=0 and m is any value between -0.7333.. and -0.1333.. You can investigate the variation of the likelihood as m and s vary by evaluating expressions like m-(-.07);s-1.0;L-plogis((V1-m)/s);2*sum(V2*log(L)+(1-V2)*log(1-L)) For instance, for any value of s0, find the value of m which maximises this. Then you can get an indication about your question by looking for the value of s such that this maximised value (with sign changed) is just on (say) the 5% point of a chisq[df=1]; my gropings suggest that s=0.8, m=(-0.1) (approx). This gives you a pair (m,s) which is just consistent with your data by this criterion. What, for instance, is the probability for any value of V1 that V2=1/0? E.g. for m=-0.1,s=0.8, consider the range -2 = x =2 (step=0.1): m-(-0.10);s-0.8;x-0.1*(-20:20);L-plogis((x-m)/s);L [1] 0.08509905 0.09534946 0.10669059 0.11920292 0.13296424 0.14804720 [7] 0.16451646 0.18242552 0.20181322 0.22270014 0.24508501 0.26894142 [13] 0.29421497 0.32082130 0.34864514 0.37754067 0.40733340 0.43782350 [19] 0.46879063 0.5000 0.53120937 0.56217650 0.5920 0.62245933 [25] 0.65135486 0.67917870 0.70578503 0.73105858 0.75491499 0.77729986 [31] 0.79818678 0.81757448 0.83548354 0.85195280 0.86703576 0.88079708 [37] 0.89330941 0.90465054 0.91490095 0.92414182 0.93245331 so that P(V2=1) can be substantial (0.1) for V1 as low as -1.8, and P(V2=0) likewise for V2 as high as +1.6; yet this (m,s) is not rejected on likelihood grounds. So, in answer to your substantive question, it would seem that your data do not support the generalisation you are asking about. And so on; you can plot things out, etc. You can do a simulation study: for a given (m,s), say the pair above, and a set of V1 values like those which you have, what is the probability that you get a set of results (V2) which show perfect separation?:-- find what proportion of simulations satisfy max(which(V1[V2==0])) min(which(V1[V2==1])) Explore a grid of (m,s) values and estimate this proportion; smooth the estimates and plot a contour diagram ... and so on! Use R as a tool for questions like this, and do not necessarily expect to find a procedure which is tailor-made for (e.g.) this particular question! Best wishes, Ted. E-Mail: (Ted Harding) [EMAIL PROTECTED] Fax-to-email: +44 (0)870 167 1972 Date: 10-Sep-03 Time: 20:24:06 -- XFMail -- __ [EMAIL PROTECTED] mailing list https://www.stat.math.ethz.ch/mailman/listinfo/r-help
RE: [R] PLS LDA
Do you mean the pls.pcr package by Prof. Wehrens? This is what I do: o Code the two groups as 0s and 1s (numeric, not factor). o Run PLS as usual. Cases with predicted values 0.5 get classified as 1s, otherwise as 0s. o Note that you need to modify the code inside the mvr() function a bit if you want to use the built-in selection of number of LVs: It selects the number that gives the best MSE, but what you really want is the number that gives the best error rate. One trick is to discretize the predictions in {0, 1}, then the MSE will be error rate. There are better ways to do this, but this works fairly well. HTH, Andy -Original Message- From: Christoph Lehmann [mailto:[EMAIL PROTECTED] Sent: Wednesday, September 10, 2003 1:38 PM To: [EMAIL PROTECTED] Subject: [R] PLS LDA Dear R experts I saw and downloaded the fresh pls package for R. Is there any way of using this pls package for PLS discriminant analysis? If not, is there any other package available. I need a way of classifying objects into e.g. two groups, where nbr_observations nbr_variables many thanks for your kind help Christoph -- Christoph Lehmann [EMAIL PROTECTED] __ [EMAIL PROTECTED] mailing list https://www.stat.math.ethz.ch/mailman/listinfo /r-help -- Notice: This e-mail message, together with any attachments,...{{dropped}} __ [EMAIL PROTECTED] mailing list https://www.stat.math.ethz.ch/mailman/listinfo/r-help
RE: [R] PLS LDA
Hi Andy Great and thanks a lot! Yes, it is the package from Prof. Wehrens. So I just run the PLS like a Logistic Regression, coding the endogenous variable as binary. So no need of specifying a binary-link function (as we have to when using glm)? And yes of course: I need the LVs which give the best error rate. What do you mean by discretize the predictions in {0, 1}? Does this mean I assign a prediction either a 0 (if predicted values =0.5) or a 1 if the predicted value is 0.5? I need to dive into the package tomorrow, so that I better understand the material, but is there any way of calculating e.g. a leaving-one-out cross-validation error? Thanks and best regards Christoph On Wed, 2003-09-10 at 21:50, Liaw, Andy wrote: Do you mean the pls.pcr package by Prof. Wehrens? This is what I do: o Code the two groups as 0s and 1s (numeric, not factor). o Run PLS as usual. Cases with predicted values 0.5 get classified as 1s, otherwise as 0s. o Note that you need to modify the code inside the mvr() function a bit if you want to use the built-in selection of number of LVs: It selects the number that gives the best MSE, but what you really want is the number that gives the best error rate. One trick is to discretize the predictions in {0, 1}, then the MSE will be error rate. There are better ways to do this, but this works fairly well. HTH, Andy -Original Message- From: Christoph Lehmann [mailto:[EMAIL PROTECTED] Sent: Wednesday, September 10, 2003 1:38 PM To: [EMAIL PROTECTED] Subject: [R] PLS LDA Dear R experts I saw and downloaded the fresh pls package for R. Is there any way of using this pls package for PLS discriminant analysis? If not, is there any other package available. I need a way of classifying objects into e.g. two groups, where nbr_observations nbr_variables many thanks for your kind help Christoph -- Christoph Lehmann [EMAIL PROTECTED] __ [EMAIL PROTECTED] mailing list https://www.stat.math.ethz.ch/mailman/listinfo /r-help -- Notice: This e-mail message, together with any attachments,...{{dropped}} __ [EMAIL PROTECTED] mailing list https://www.stat.math.ethz.ch/mailman/listinfo/r-help -- Christoph Lehmann [EMAIL PROTECTED] __ [EMAIL PROTECTED] mailing list https://www.stat.math.ethz.ch/mailman/listinfo/r-help
[R] coef names in lm
Dear all, I am interested in finding out how to change the names of coefficients in the lm function. I have a design matrix which I called design where each variate has its own name. However when I issue the command: lm.1-lm(response~design-1, weights=some.weights) and follow it with: summary(lm.1) it seems to paste as a character the names of the variates with design i.e I have something like: designAge designPlace designOccupation ... as names of coefficients and instead I just wanted to be it seems to do Age Place Occupation. P.S.. the reason I need this is because I am using the xtable library to turn output into latex tables and do not want to manually delete each and every single design word in the coefficients name vector. Thank you so much for any feedback, Jean Eid __ [EMAIL PROTECTED] mailing list https://www.stat.math.ethz.ch/mailman/listinfo/r-help
RE: [R] coef names in lm
If you coerce design into a data frame and then do lm.1 - lm(response ~ . - 1, data=design, ...) that should work. Andy -Original Message- From: Jean Eid [mailto:[EMAIL PROTECTED] Sent: Wednesday, September 10, 2003 4:28 PM To: [EMAIL PROTECTED] Subject: [R] coef names in lm Dear all, I am interested in finding out how to change the names of coefficients in the lm function. I have a design matrix which I called design where each variate has its own name. However when I issue the command: lm.1-lm(response~design-1, weights=some.weights) and follow it with: summary(lm.1) it seems to paste as a character the names of the variates with design i.e I have something like: designAge designPlace designOccupation ... as names of coefficients and instead I just wanted to be it seems to do Age Place Occupation. P.S.. the reason I need this is because I am using the xtable library to turn output into latex tables and do not want to manually delete each and every single design word in the coefficients name vector. Thank you so much for any feedback, Jean Eid __ [EMAIL PROTECTED] mailing list https://www.stat.math.ethz.ch/mailman/listinfo /r-help -- Notice: This e-mail message, together with any attachments,...{{dropped}} __ [EMAIL PROTECTED] mailing list https://www.stat.math.ethz.ch/mailman/listinfo/r-help
Re: [R] coef names in lm
Maybe this will work for you: df - as.data.frame(design) lm.2 - lm(response ~ ., df) -roger Jean Eid wrote: Dear all, I am interested in finding out how to change the names of coefficients in the lm function. I have a design matrix which I called design where each variate has its own name. However when I issue the command: lm.1-lm(response~design-1, weights=some.weights) and follow it with: summary(lm.1) it seems to paste as a character the names of the variates with design i.e I have something like: designAge designPlace designOccupation ... as names of coefficients and instead I just wanted to be it seems to do Age Place Occupation. P.S.. the reason I need this is because I am using the xtable library to turn output into latex tables and do not want to manually delete each and every single design word in the coefficients name vector. Thank you so much for any feedback, Jean Eid __ [EMAIL PROTECTED] mailing list https://www.stat.math.ethz.ch/mailman/listinfo/r-help -- Together, we can stop attaching Word documents http://www.fsf.org/philosophy/no-word-attachments.html __ [EMAIL PROTECTED] mailing list https://www.stat.math.ethz.ch/mailman/listinfo/r-help
[R] regression questions
I have been puzzling over how to fit some fixed effects models to a set of data. My response function is response - function(a, b, c, alpha1, alpha2, indicator, t, t2) { z = a + b * (t) * exp(-alpha1 * t) + indicator *c * (t2) * exp(-alpha2 * t2) } where t2 = t - 4 and indicator is a 0-1 vector denoting when t 4. Each test subject receives equal doses at t = 0 and t = 4. The dose can vary from subject to subject. Also note the following: 1. Var(e[it]) = sigma1^2 for t=4; Var(e[it]) = sigma2^2 for t4. This is motivated by my data exploration. 2. b,c 0 for biological interpretability 3. t varies over {0,2,4,6,8,10}. 4. For a variety of reasons, a, alpha1, and alpha2 must be held constant over all of the test subjects. The function nlsList( ) is not appropriate because it assumes that all of the parameters are allowed to vary with each level of a specified grouping variable (in this case, subject.id). I have been able to fit nls( ) models using the following syntax: model.nls1 - nls(y ~ response(10, b[subject.id], c[subject.id], alpha1, alpha2, indicator, t, t2), data = foo.frame, start = list(b = rep(25,12), c = rep(100,12), alpha1 = 0.5, alpha2 = 0.5), trace = T) The start values were motivated by some data exploration, and the results appear to be stable. The value a=10 was fixed also as a result of the initial data exploration, and appears necessary in order for the model to be stable. Unfortunately, the estimated b- and c-values for several subjects are negative. Also, nls( ) does not allow a weights = statement like gnls( ) does. When I try model.nls1 - gnls(y ~ response(10, b[subject.id], c[subject.id], alpha1, alpha2, indicator, t, t2), data = foo.frame, start = list(b = rep(25,12), c = rep(100,12), alpha1 = 0.5, alpha2 = 0.5), trace = T) I get the message Error in eval(expr, envir, enclos) : Object b not found This surprises me, since my understanding is that gnls( ) is essentially nls( ) but with weights = and correlation = options. I suppose that separate fixed effects for each subject could be estimated from gnls( ) if I created a separate indicator variable for each subject and added them to the data frame (I have not yet done this); however, this does not address the need for the b,c parameters to be constrained greater than zero. I would gratefully welcome suggestions. Much thanks in advance, david paul __ [EMAIL PROTECTED] mailing list https://www.stat.math.ethz.ch/mailman/listinfo/r-help
Re: [R] Computing a CDF or many quantiles
Your method looks like a naive reimplementation of integration, and won't work so well for distributions that have the great majority of the probability mass concentrated in a small fraction of the sample space. I was hoping for something that would retain the adaptability of integrate(). (Ted Harding) wrote: If that's all you want to do, then a very straightfoward approach should be OK. I illustrate with a truncated normal distribution on [-1,1]: x - (-1)+(0.001*(0:2000));pdf-dnorm(x); pdf-pdf/(sum(pdf)*0.001) CDF-cumsum(pdf)*0.001 plot(x,pdf,ylim=c(0,1),type=l);lines(x,CDF) Quantiles: N=10;e-CDF[1]; for(i in (0:10)){ j-max(which(CDF=i/N+e));print(c(x[j],CDF[j])) } __ [EMAIL PROTECTED] mailing list https://www.stat.math.ethz.ch/mailman/listinfo/r-help
Re: [R] Computing a CDF or many quantiles
On September 10, 2003 04:03 pm, Kevin S. Van Horn wrote: Your method looks like a naive reimplementation of integration, and won't work so well for distributions that have the great majority of the probability mass concentrated in a small fraction of the sample space. I was hoping for something that would retain the adaptability of integrate(). Yesterday, I've suggested to use approxfun(). Did you consider my suggestion? Below is an example. N - 500 x - rexp(N) y - rank(x)/(N+1) empCDF - approxfun(x,y) xvals - seq(0,4,.01) plot(xvals,empCDF(xvals),type=l, xlab=Quantile,ylab=Cumulative Distribution Function) lines(xvals,pexp(xvals),lty=2) legend(2,.4,c(Empirical CDF,Exact CDF),lty=1:2) It's possible to tune in some parameters in approxfun() to better match your personal preferences. Have a look at help(approxfun) for details. HTH, Jerome Asselin __ [EMAIL PROTECTED] mailing list https://www.stat.math.ethz.ch/mailman/listinfo/r-help
Re: [R] Computing a CDF or many quantiles
Also look at ecdf() from package stepfun. HTH, Jerome Asselin __ [EMAIL PROTECTED] mailing list https://www.stat.math.ethz.ch/mailman/listinfo/r-help
[R] Customised legend in lattice
Hi List, Am trying to customize a legend in trellis: Draws 2x5 lines in 5 colors and 2 linetypes. I would like to add two more items to the legend showing the key for the line types above the colored legend. Any suggestions welcome - thanks Herry # #Following example code: library(gregmisc) trellis.device(bg=white) i1=0 i2=-1.89767506 i3=-1.17087085 i4=-0.09853587 i5=0.87486320 a1=1.9106842 a2=0.7518519 x1-seq(0,7, by=0.01) y1-inv.logit(a1+a2*(x1)+i1) y2-inv.logit(a1+a2*(x1)+i2) y3-inv.logit(a1+a2*(x1)+i3) y4-inv.logit(a1+a2*(x1)+i4) y5-inv.logit(a1+a2*(x1)+i5) z1-inv.logit(a2*(x1)+i1) z2-inv.logit(a2*(x1)+i2) z3-inv.logit(a2*(x1)+i3) z4-inv.logit(a2*(x1)+i4) z5-inv.logit(a2*(x1)+i5) as.data.frame(cbind(y1,y2,y3,y4,y5,z1,z2,z3,z4,z5))-tmp leg.txt-c(Rating 1,Rating 2,Rating 3,Rating 4,Rating 5) colo-rep(c(black,red,darkgreen,navyblue,rosybrown),2) plot4-xyplot(y1+y2+y3+y4+y5+z1+z2+z3+z4+z5~x1, cex=2, xlab=number x, ylab=Probability, bg=white, xlim=c(0,7), ylim=c(0,1), type=l, allow.multiple=TRUE, data=tmp, panel=function(x,y,subscripts,...){ for(i in 1:5){ panel.xyplot(x1,tmp[,i],col=colo[i],lwd=2,lty=1,...) } for(i in 6:10){ panel.xyplot(x1,tmp[,i],col=colo[i],lwd=2,lty=2,...) } } ) update(plot4, key = list(corner=c(0,1), x=0.65, y=0.35, points=list(c(1:5),col=colo[1:5],pch=19), text=list(leg.txt), ) ) ##--- Not working part--XXX update(plot4, key = list(corner=c(0,1), x=0.65, y=0.35, lines=list(c(1:2),col=black,lwd=1,lty=c(1:2)), text=list(c(category 1,category 2)), points=list(c(1:5),col=colo[1:5],pch=19), text=list(leg.txt), ) ) Alexander Herr - Herry Northern Futures Davies Laboratory PMB, Aitkenvale, QLD 4814 Phone (07) 4753 8510 Fax (07) 4753 8650 Home: http://batcall.csu.edu.au/~aherr Webadmin ABS: http://ausbats.org.au Sustainable Ecosystems: http://www.cse.csiro.au/ [[alternative HTML version deleted]] __ [EMAIL PROTECTED] mailing list https://www.stat.math.ethz.ch/mailman/listinfo/r-help
Re: [R] scan() problem
If the records are always of the form: number,...,...,number where ... may contain commas but not double quotes then here is a kludgy solution. Perhaps its sufficient? # scan in data using as the delimiter and keep first and last fields s - scan(clipboard,skip=1,what=list(,NULL,NULL,NULL,),sep=\) # remove commas from fields, convert to numeric and reshape into matrix matrix(as.numeric(sub(,,,unlist(s))),nc=2) --- Paul Bayer [EMAIL PROTECTED] wrote: Dear R-helpers, I have to read some large csv-files into R (30 - 100MB). Since reading with read.csv leads to memory exhausted, I tried with scan(), skipping not needed columns by NULL-elements in what. When these skipped elements are quoted strings with commata inside, R interprets each such quoted comma as element separator leading to wrong records in the rest of the line. A little test will show what I mean. I have the following test.csv: col.A,col.B,col.C,col.D 1,quoted string,again, again again,123 2,nice quotes, isnt it,you got it,456 First I read all elements: tst - scan(test.csv, what=list(a=0,b=,c=,d=0), sep=,, skip=1) Read 2 records tst $a [1] 1 2 $b [1] quoted stringnice quotes, isnt it $c [1] again, again again you got it $d [1] 123 456 Everything is fine. Then I try to skip the 2nd column by giving b=NULL: tst - scan(test.csv, what=list(a=0,b=NULL,c=,d=0), sep=,, skip=1) Read 2 records Warning message: number of items read is not a multiple of the number of columns tst $a [1] 1 2 $b NULL $c [1] again, again again isnt it,you got it,456\n\n\n $d [1] 123 NA I got garbage. Isn't this a bug? Or did I something wrong? Is there a workaround? Thank you all, Paul Bayer, Feldafing, Germany __ [EMAIL PROTECTED] mailing list https://www.stat.math.ethz.ch/mailman/listinfo/r-help __ [EMAIL PROTECTED] mailing list https://www.stat.math.ethz.ch/mailman/listinfo/r-help