Re: [R] String substitution
Wonderful! Thank you Arun! Irene  Irene Ruberto Da: arun kirshna [via R] ml-node+s789695n4677558...@n4.nabble.com Inviato: Giovedì 3 Ottobre 2013 22:51 Oggetto: Re: String substitution Hi, Try: dat$y- as.character(dat$y) dat1- dat dat2- dat library(stringr)  dat$y[NET]- substr(word(dat$y[NET],2),1,1)  dat$y #[1] n    n    house n    tree #or for(i in 1:length(NET)){dat1$y[NET[i]]- n}  dat1$y #[1] n    n    house n    tree #or dat2$y[NET]- gsub(.*(n).*,\\1,dat2$y[NET])  dat2$y #[1] n    n    house n    tree A.K. Hello, I am trying to replace strings containing a certain word, I first identified the word (in this example net) with grep, and then I need to replace those string with n. It should be very simple but I don't seem to find the solution.  Example: x-c(5:9) y- c(with net, with nets, house, no nets, tree) dat-as.data.frame(cbind(x, y) ) NET-grep(net, dat$y) # I want y to become (n, n, house, n, tree) # # I have tried several ways including the following but without success # for (i in 1: length(NET)) { dat$y[NET[i]]- n } Thank you for your help! __ [hidden email] 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. If you reply to this email, your message will be added to the discussion below:http://r.789695.n4.nabble.com/String-substitution-tp4677541p4677558.html To unsubscribe from String substitution, click here. NAML -- View this message in context: http://r.789695.n4.nabble.com/String-substitution-tp4677541p4677612.html Sent from the R help mailing list archive at Nabble.com. [[alternative HTML version deleted]] __ R-help@r-project.org mailing list https://stat.ethz.ch/mailman/listinfo/r-help PLEASE do read the posting guide http://www.R-project.org/posting-guide.html and provide commented, minimal, self-contained, reproducible code.
[R] help: R GUI front-end has stopped working
Hello, I'm using the function nlminb of the package stats inside a loop and when the number of trials grows, R crashes and says R GUI front-end has stopped working. Could you help me with this problem? I have try in versions 2.15.1,2.15.2 and 3.0.0. sessionInfo() R version 2.15.2 (2012-10-26) Platform: i386-w64-mingw32/i386 (32-bit) locale: [1] LC_COLLATE=Galician_Spain.1252 LC_CTYPE=Galician_Spain.1252 [3] LC_MONETARY=Galician_Spain.1252 LC_NUMERIC=C [5] LC_TIME=Galician_Spain.1252 attached base packages: [1] stats graphics grDevices utils datasets methods base This is my code: u-runif(3000) k=13 n=length(u) prob=vector(length=n) ro=vector(length=n) beta1=vector(length=n) g1=vector(length=n) sb1=vector(length=n) low=vector(length=n) tlow=vector(length=n) effects=vector(length=n) vb1=vector(length=n) AA=list() BB=list() nnI=list() library(stats) for (i in 1:n){ v=as.numeric(u=u[i]) p=mean(v) A=vector(length=k) n1=n%/%k A-sapply(1:k,function(m) A-sum(v[((m-1)*n1+1):(m*n1)])) A[k]=sum(v[((k-1)*n1+1):n]) AA[[i]]-A B=c(rep(n1,k-1),length(v[((k-1)*n1+1):n])) BB[[i]]-B l1=vector(length=(length(A))) L1-function(pe,rho){ for(j in 1:length(A)){ if(A[j]==0){l1[j]=0} else {l1[j]-sum(log(pe+(-rho/(rho-1))*(0:(A[j]-1} } return(sum(l1))} l2=vector(length=(length(B-A))) L2-function(pe,rho){ for(j in 1:length(B-A)){ if((B[j]-A[j])==0){l2[j]=0} else {l2[j]-sum(log(1-pe+(-rho/(rho-1))*(0:(B[j]-A[j]-1} } return(sum(l2))} l3=vector(length=(length(B))) L3-function(pe,rho){ for(j in 1:length(A)){ if(B[j]==0){l1[j]=0} else {l3[j]-sum(log(1+(-rho/(rho-1))*(0:(B[j]-1} } return(sum(l3))} L-function(pe,rho){ L-L1(pe,rho)+L2(pe,rho)-L3(pe,rho) return(L) } Max- function(x){ -L(x[1], x[2])} opt-nlminb(c(0.01,0.01), Max,lower = rep(0.001,2), upper = rep(0.999,2),control=list(rel.tol=1e-6)) prob[i]=opt$par[1] ro[i]=opt$par[2] } Thanks, Irene Castro Conde. [[alternative HTML version deleted]] __ R-help@r-project.org mailing list https://stat.ethz.ch/mailman/listinfo/r-help PLEASE do read the posting guide http://www.R-project.org/posting-guide.html and provide commented, minimal, self-contained, reproducible code.
Re: [R] long margin text below lattice plot - how to wrap lines?
Thanks a million! - Original Message - From: Pascal Oettli Sent: 12/12/12 09:37 AM To: knallg...@gmx.com Subject: Re: [R] long margin text below lattice plot - how to wrap lines? Hello You can insert \n in your text. mytext - textGrob(This is such a very very long text\n that it goes on forever and therefore needs to be wrapped in order\n for someone to be able to read it properly.) HTH Pascal Le 12/12/2012 16:22, knallg...@gmx.com a écrit : Hello, I've got a lattice plot and need to add text into the bottom margin of the plotting area (below the bottom legend). This seems to work in principle using grid.arrange, yet the text to be added is rather long. As a consequence, it gets clipped: require(lattice) require(grid) myplot - xyplot(1~1) mytext - textGrob(This is such a very very long text that it goes on forever and therefore needs to be wrapped in order for someone to be able to read it properly.) grid.arrange(myplot, sub=mytext) Is there any way to wrap the lines of the text or some other of workaround to this problem? I've tried figuring it out myself, but it's a rather urgent problem and therefore I was hoping to find some help here. Best, Irene __ R-help@r-project.org mailing list https://stat.ethz.ch/mailman/listinfo/r-help PLEASE do read the posting guide http://www.R-project.org/posting-guide.html and provide commented, minimal, self-contained, reproducible code. __ R-help@r-project.org mailing list https://stat.ethz.ch/mailman/listinfo/r-help PLEASE do read the posting guide http://www.R-project.org/posting-guide.html and provide commented, minimal, self-contained, reproducible code.
[R] match and substitute two variables
Hello, I have two variables (of different length and from two different data frames): code- c(101001, 1032, 102, 101001, 102, 1032); name- c(101001 Alta, 102 Bassa, 1032 Media); and I would like to substitute the first variable with the second variable according to their shared numerical part, thus obtaining the following result: code.new 101001 Alta 1032 Media 102 Bassa101001 Alta 102 Bassa 1032 Media I tried using: - sapply(code, gsub, pattern=\\d+, replacement=name) but the replacement cannot be of length more than one, thus my output is only 101001 Alta 101001 Alta... I am not sure how to get the right answer... Thank you! -- View this message in context: http://r.789695.n4.nabble.com/match-and-substitute-two-variables-tp4651893.html Sent from the R help mailing list archive at Nabble.com. __ R-help@r-project.org mailing list https://stat.ethz.ch/mailman/listinfo/r-help PLEASE do read the posting guide http://www.R-project.org/posting-guide.html and provide commented, minimal, self-contained, reproducible code.
Re: [R] match and substitute two variables
It works perfectly, thank you! -- View this message in context: http://r.789695.n4.nabble.com/match-and-substitute-two-variables-tp4651893p4651906.html Sent from the R help mailing list archive at Nabble.com. __ R-help@r-project.org mailing list https://stat.ethz.ch/mailman/listinfo/r-help PLEASE do read the posting guide http://www.R-project.org/posting-guide.html and provide commented, minimal, self-contained, reproducible code.
[R] Extracting week number starting from a specific date
Hello, I have a list of dates, such as dates- as.Date(c(1996-03-29,1996-05-30,1996-09-28,1996-05-09)) from which I would like to extract the week number for each date, with week n°1 being the week going from Dec 30th 1995 to Jan 6th 1996 (1995-30-12 to 1996-06-01). Any suggestion for a simple way to do that? Thank you -- View this message in context: http://r.789695.n4.nabble.com/Extracting-week-number-starting-from-a-specific-date-tp4598676.html Sent from the R help mailing list archive at Nabble.com. __ R-help@r-project.org mailing list https://stat.ethz.ch/mailman/listinfo/r-help PLEASE do read the posting guide http://www.R-project.org/posting-guide.html and provide commented, minimal, self-contained, reproducible code.
Re: [R] Extracting numbers from a character variable of different types
It worked perfectly! Thank you -- View this message in context: http://r.789695.n4.nabble.com/Extracting-numbers-from-a-character-variable-of-different-types-tp4482248p4505914.html Sent from the R help mailing list archive at Nabble.com. __ R-help@r-project.org mailing list https://stat.ethz.ch/mailman/listinfo/r-help PLEASE do read the posting guide http://www.R-project.org/posting-guide.html and provide commented, minimal, self-contained, reproducible code.
[R] Extracting numbers from a character variable of different types
Hello, I have a file which contains a column with age, which is represented in the two following patterns 1. 007/A or ''007/a or ''7 /a . In this case A or a means year and I would like to extract only the numeric values eg 7 in the above case if this pattern exits in a line of file. 2. 004/M or 004/m where M or m means month .. for these lines I would like to first extract the numeric value of Month eg. 4 and then convert it into a value of years, which would be 0.33 eg 4 divided by 12. Can anyone help? Thank you -- View this message in context: http://r.789695.n4.nabble.com/Extracting-numbers-from-a-character-variable-of-different-types-tp4482248p4482248.html Sent from the R help mailing list archive at Nabble.com. __ R-help@r-project.org mailing list https://stat.ethz.ch/mailman/listinfo/r-help PLEASE do read the posting guide http://www.R-project.org/posting-guide.html and provide commented, minimal, self-contained, reproducible code.
[R] color of error bars in Dotplot (Hmisc)
Hello! In a grouped Dotplot, is there any way to set the color of error bars to be the same as the corresponding symbols? Example data: require(lattice) require(Hmisc) data(barley) Dotplot(variety~Cbind(yield, yield+2, yield-2)|year, groups=site, data=barley) I experimented with changing trellis settings of plot.line (as mentioned in the Hmisc documentation) as well as col.line settings in simpleTheme, but that didn't work. Changing the error bars to any one other color does work, but I can't seem to figure out how to set them to the color of each group-specific symbol... I'd be grateful for any advice, Irene __ R-help@r-project.org mailing list https://stat.ethz.ch/mailman/listinfo/r-help PLEASE do read the posting guide http://www.R-project.org/posting-guide.html and provide commented, minimal, self-contained, reproducible code.
[R] fitting gamm with interaction term
Hi all, I would like to fit a gamm model of the form: Y~X+X*f(z) Where f is the smooth function and With random effects on X and on the intercept. So, I try to write it like this: gam.lme- gamm(Y~ s(z, by=X) +X, random=list(groups=pdDiag(~1+X)) ) but I get the error message : Error in MEestimate(lmeSt, grps) : Singularity in backsolve at level 0, block 1 When I simply fit a gam model using the formula above, then it works ok. Is it possible to fit such a model with gamm? Thanks a lot! [[alternative HTML version deleted]] __ R-help@r-project.org mailing list https://stat.ethz.ch/mailman/listinfo/r-help PLEASE do read the posting guide http://www.R-project.org/posting-guide.html and provide commented, minimal, self-contained, reproducible code.
[R] fitting gamm with interaction term
Hi all, I would like to fit a gamm model of the form: Y~X+X*f(z) Where f is the smooth function and With random effects on X and on the intercept. So, I try to write it like this: gam.lme- gamm(Y~ s(z, by=X) +X, random=list(groups=pdDiag(~1+X)) ) but I get the error message : Error in MEestimate(lmeSt, grps) : Singularity in backsolve at level 0, block 1 When I simply fit a gam model using the formula above, then it works ok. Is it possible to fit such a model with gamm? Thanks a lot! [[alternative HTML version deleted]] __ R-help@r-project.org mailing list https://stat.ethz.ch/mailman/listinfo/r-help PLEASE do read the posting guide http://www.R-project.org/posting-guide.html and provide commented, minimal, self-contained, reproducible code.
[R] Conditional editing of rows in a data frame
Dear R users, I have a dataframe (main.table) with ~30,000 rows and 6 columns, of which here are a few rows: id chr window gene xp.normxp.top 129 1_32 1 32 TAS1R1 1.28882115 FALSE 130 1_32 1 32 ZBTB48 1.28882115 FALSE 131 1_32 1 32 KLHL21 1.28882115 FALSE 132 1_32 1 32PHF13 1.28882115 FALSE 133 1_33 1 33PHF13 1.02727430 FALSE 134 1_33 1 33THAP3 1.02727430 FALSE 135 1_33 1 33 DNAJC11 1.02727430 FALSE 136 1_33 1 33 CAMTA1 1.02727430 FALSE 137 1_34 1 34 CAMTA1 1.40312732 TRUE 138 1_35 1 35 CAMTA1 1.52104538 FALSE 139 1_36 1 36 CAMTA1 1.04853732 FALSE 140 1_37 1 37 CAMTA1 0.64794094 FALSE 141 1_38 1 38 CAMTA1 1.23026086 TRUE 142 1_38 1 38VAMP3 1.23026086 TRUE 143 1_38 1 38 PER3 1.23026086 TRUE 144 1_39 1 39 PER3 1.18154967 TRUE 145 1_39 1 39 UTS2 1.18154967 TRUE 146 1_39 1 39 TNFRSF9 1.18154967 TRUE 147 1_39 1 39PARK7 1.18154967 TRUE 148 1_39 1 39 ERRFI1 1.18154967 TRUE 149 1_40 1 40 no_gene 1.79796879 FALSE 150 1_41 1 41 SLC45A1 0.20193560 FALSE I want to create two new columns, xp.bg and xp.n.top, using the following criteria: If gene is the same in consecutive rows, xp.bg is the minimum value of xp.norm in those rows; if gene is not the same, xp.bg is simply the value of xp.norm for that row; Likewise, if there's a run of contiguous xp.top = TRUE values, xp.n.top is the minimum value in that range, and if xp.top is false or NA, xp.n.top is NA, or 0 (I don't care). So, in the above example, xp.bg for rows 136:141 should be 0.64794094, and is equal to xp.norm for all other rows, xp.n.top for row 137 is 1.40312732, 1.18154967 for rows 141:148, and 0/NA for all other rows. Is there a way to combine indexing and if statements or some such to accomplish this? I want to it this without using split(main.table, main.table$gene), because there's about 20,000 unique entries for gene, and one of the entries, no_gene, is repeated throughout. I thought briefly of subsetting the rows where xp.top is TRUE, but I then don't know how to set the range for min, so that it only looks at what would originally have been consecutive rows, and searching the help has not proved particularly useful. Thanks in advance, Irene Gallego Romero -- Irene Gallego Romero Leverhulme Centre for Human Evolutionary Studies University of Cambridge Fitzwilliam St Cambridge CB1 3QH UK email: ig...@cam.ac.uk __ R-help@r-project.org mailing list https://stat.ethz.ch/mailman/listinfo/r-help PLEASE do read the posting guide http://www.R-project.org/posting-guide.html and provide commented, minimal, self-contained, reproducible code.
[R] Scientific Format E
Dear Helpers, I would like to export a large dataset to a txt file in order to use it in an other programm. Unfurtunatly the R the scientific format is a small e: 2 e-1 while the other programm requires the format to be a capital E: 2E-1 How can I change this in R? Thanks for your help PS: I already found the command to turn them into a non scientific format (scientific=FALSE) [[alternative HTML version deleted]] __ R-help@r-project.org mailing list https://stat.ethz.ch/mailman/listinfo/r-help PLEASE do read the posting guide http://www.R-project.org/posting-guide.html and provide commented, minimal, self-contained, reproducible code.
[R] How estimate VAR(p)-model robustly?
Hello, Does anyone know about robust estimation of vector autoregressive models (VAR(p)) in R? Or in Matlab? Currently I am using the function ar(). The problem is, that the variances of my data change a lot with time, and we also have some outliers in the data. That is why, I presume, that we would get quite different results when estimating robustly. I would be very grateful if someone could help! Thanks a lot! Irene. [[alternative HTML version deleted]] __ R-help@r-project.org mailing list https://stat.ethz.ch/mailman/listinfo/r-help PLEASE do read the posting guide http://www.R-project.org/posting-guide.html and provide commented, minimal, self-contained, reproducible code.
[R] ARMA-GARCH package in R?
Hello, Does anyone know about an R-package on multivariate ARMA-GARCH models? Or in Matlab? I would be very grateful if someone could help! Thanks a lot! [[alternative HTML version deleted]] __ R-help@r-project.org mailing list https://stat.ethz.ch/mailman/listinfo/r-help PLEASE do read the posting guide http://www.R-project.org/posting-guide.html and provide commented, minimal, self-contained, reproducible code.
[R] Sliding window over irregular intervals
Dear all, I have some very big data files that look something like this: id chr pos ihh1 ihh2 xpehh rs5748748 22 15795572 0.0230222 0.0268394 -0.153413 rs5748755 22 15806401 0.0186084 0.0268672 -0.367296 rs2385785 22 15807037 0.0198204 0.0186616 0.0602451 rs1981707 22 15809384 0.0299685 0.0176768 0.527892 rs1981708 22 15809434 0.0305465 0.0187227 0.489512 rs11914222 22 15810040 0.0307183 0.0172399 0.577633 rs4819923 22 15813210 0.02707 0.0159736 0.527491 rs5994105 22 15813888 0.025202 0.0141296 0.578651 rs5748760 22 15814084 0.0242894 0.0146486 0.505691 rs2385786 22 15816846 0.0173057 0.0107816 0.473199 rs1990483 22 15817310 0.0176641 0.0130525 0.302555 rs5994110 22 15821524 0.0178411 0.0129001 0.324267 rs17733785 22 15822154 0.0201797 0.0182093 0.102746 rs7287116 22 15823131 0.0201993 0.0179028 0.12069 rs5748765 22 15825502 0.0193195 0.0176513 0.090302 I'm trying to extract the maximum and minimum xpehh (last column) values within a sliding window (non overlapping), of width 1 (calculated relative to pos (third column)). However, as you can tell from the brief excerpt here, although all possible intervals will probably be covered by at least one data point, the number of data points will be variable (incidentally, if anyone knows of a way to obtain this number, that would be lovely), as will the spacing between them. Furthermore, values of chr (second column) will range from 1 to 22, and values of pos will be overlapping across them; I want to evaluate the window separately for each value of chr. I've looked at the help and FAQ on sliding windows, but I'm a relative newcomer to R and cannot find a way to do what I need to do. Everything I've managed to unearth so far seems geared towards smoother time series. Any help on this problem would be vastly appreciated. Thanks, Irene -- Irene Gallego Romero Leverhulme Centre for Human Evolutionary Studies University of Cambridge Fitzwilliam St Cambridge CB2 1QH UK __ R-help@r-project.org mailing list https://stat.ethz.ch/mailman/listinfo/r-help PLEASE do read the posting guide http://www.R-project.org/posting-guide.html and provide commented, minimal, self-contained, reproducible code.
[R] package ccgarch - dcc.estimation
Hello, I am trying to model a bivariate time series called 'residuals' as a dcc-garch model. I want to use the function dcc.estimation(a, A, B dcc.para, dvar, model) out of the package ccgarch to estimate the parameters. No matter how I tried to define a, A and B, I always got the message Error in constrOptim(theta = para, f = loglik.dcc2, gr = grad.dcc2, ui = resta, : initial value not feasible. As my knowledge about GARCH is rather poor, I might have defined some input values in an incorrect way. E.g. this is a part of my code: a - rep(0,2) A - diag(2) B - diag(2) dcc.para - rep(0,2) dvar - residuals model - 'diagonal' dcc.estimation(a, A, B, dcc.para, dvar, model) I would be very grateful if someone could help! Thanks a lot! [[alternative HTML version deleted]] __ R-help@r-project.org mailing list https://stat.ethz.ch/mailman/listinfo/r-help PLEASE do read the posting guide http://www.R-project.org/posting-guide.html and provide commented, minimal, self-contained, reproducible code.
[R] package ccgarch - dcc.estimation
Hello, I am trying to model a bivariate time series called 'residuals' as a dcc-garch model. I want to use the function dcc.estimation(a, A, B dcc.para, dvar, model) to estimate the parameters. No matter how I tried to define a, A and B, I always got the message Error in constrOptim(theta = para, f = loglik.dcc2, gr = grad.dcc2, ui = resta, : initial value not feasible. As my knowledge about GARCH is rather poor, I might have defined some input values in an incorrect way. E.g. this is a part of my code: a - rep(0,2) A - diag(2) B - diag(2) dcc.para - rep(0,2) dvar - residuals model - 'diagonal' dcc.estimation(a, A, B, dcc.para, dvar, model) I would be very grateful if someone could help! Thanks a lot! [[alternative HTML version deleted]] __ R-help@r-project.org mailing list https://stat.ethz.ch/mailman/listinfo/r-help PLEASE do read the posting guide http://www.R-project.org/posting-guide.html and provide commented, minimal, self-contained, reproducible code.
[R] preprocessCore
Bonjour! Je suis en train de faire un projet utilisant Affymatrix mais j'ai un problème à télécharger un package depuis R: une fois que je télécharge le package et je demande de faire rma ca sort l'erreur suivante et je n'ai aucune idée de comment je pourrais faire. library(preprocessCore) plac.rma - rma(plac.new) Background correcting Normalizing Calculating Expression Errore in rma(plac.new) : function 'R_subColSummarize_medianpolish_log' not provided by package 'preprocessCore' Est-ce que vous avez un conseil à me donner? J'ai un projet à finir et je suis complètement bloquée :( Merci d'avance, Irene Vicari __ R-help@r-project.org mailing list https://stat.ethz.ch/mailman/listinfo/r-help PLEASE do read the posting guide http://www.R-project.org/posting-guide.html and provide commented, minimal, self-contained, reproducible code.
[R] Confidence intervals for non-lm curve
Hi all! I would like to estimate confidence intervals for a non lm model. For example, I use a mixed model of the form: md=lme(y~x1+I(x1^2)+x2 ...) Parameters x1+I(x1^2) are fixed effects and I would like to plot the predicted (partial) curve corresponding to these ones, along with 90% CI bands. Thus, I simulate (partial) predictions: Lx=c() Ux=c() Mx=c() z=c() for (j in 1:length(x1)){ x=x1[j] for(i in 1:2000){ s1 - rnorm(1,.026,.027) # mean and sd estimated by lme for x1 s2 - rnorm(1,-.01,.005) # mean and sd estimated by lme for I(x1^2) z[i] - s1*x+s2*(x^2) } Lx[j]=quantile(z,.05) Ux[j]=quantile(z,.95) Mx[j]=mean(z) } And then plot vectors Lx, Mx and Ux for lower, mean and upper curves, respectively. Is this approach correct? Any alternatives? Thank you!! __ R-help@r-project.org mailing list https://stat.ethz.ch/mailman/listinfo/r-help PLEASE do read the posting guide http://www.R-project.org/posting-guide.html and provide commented, minimal, self-contained, reproducible code.
[R] Confidence intervals for non-lm curve
Hi all! I would like to estimate confidence intervals for a non lm model. For example, I use a mixed model of the form: md=lme(y~x1+I(x1^2)+x2 ...) Parameters x1+I(x1^2) are fixed effects and I would like to plot the predicted (partial) curve corresponding to these ones, along with 90% CI bands. Thus, I simulate (partial) predictions: Lx=c() Ux=c() Mx=c() z=c() for (j in 1:length(x1)){ x=x1[j] for(i in 1:2000){ s1 - rnorm(1,.026,.027) # mean and sd estimated by lme for x1 s2 - rnorm(1,-.01,.005) # mean and sd estimated by lme for I(x1^2) z[i] - s1*x+s2*(x^2) } Lx[j]=quantile(z,.05) Ux[j]=quantile(z,.95) Mx[j]=mean(z) } And then plot vectors Lx, Mx and Ux for lower, mean and upper curves, respectively. Is this approach correct? Any alternatives? Thank you!! __ R-help@r-project.org mailing list https://stat.ethz.ch/mailman/listinfo/r-help PLEASE do read the posting guide http://www.R-project.org/posting-guide.html and provide commented, minimal, self-contained, reproducible code.
[R] acf function
Dear all, I have an annual time-series of population numbers and I would like to estimate the auto-correlation. Can I use acf() function and judge whether auto-correlation is significant by the plots? The acf array, eg: Autocorrelations of series 'x$log.s.r', by lag 0 1 2 3 4 5 6 7 8 9 10 11 12 1.000 0.031 -0.171 -0.451 0.021 0.070 0.238 -0.079 -0.046 0.006 0.188 -0.134 -0.016 13 14 15 -0.153 0.146 0.096 gives the auto-correlation at lags 1, 2 Is that right? Thank you! [[alternative HTML version deleted]] __ R-help@r-project.org mailing list https://stat.ethz.ch/mailman/listinfo/r-help PLEASE do read the posting guide http://www.R-project.org/posting-guide.html and provide commented, minimal, self-contained, reproducible code.
[R] acf function
Dear all, I have an annual time-series of population numbers and I would like to estimate the auto-correlation. Can I use acf() function and judge whether auto-correlation is significant by the plots? The acf array produced by this functions gives the auto-correlation at lags 1, 2 Is that right? Thank you! [[alternative HTML version deleted]] __ R-help@r-project.org mailing list https://stat.ethz.ch/mailman/listinfo/r-help PLEASE do read the posting guide http://www.R-project.org/posting-guide.html and provide commented, minimal, self-contained, reproducible code.
[R] points size in plots
Dear list, I would like to produce a plot of variables where the size of the points will be indicative of their standard errors. How is that possible? Thank you! [[alternative HTML version deleted]] __ R-help@r-project.org mailing list https://stat.ethz.ch/mailman/listinfo/r-help PLEASE do read the posting guide http://www.R-project.org/posting-guide.html and provide commented, minimal, self-contained, reproducible code.
[R] partial residuals
Dear all, I am trying to estimate partial residuals for the multiple regression lm model: a.lm=lm(y~x1+x2) I use the function residuals(a.lm, type=partial) However, the results are much different when I use the manual method to get partial residuals for x2 (or for x1): residuals(a.lm) +b1*x1 Where b1 is the estimated coefficient for x1 in the lm model and x1 is the vector of the first predictor. What is the difference (or my error) in these methods? Thank you! [[alternative HTML version deleted]] __ R-help@r-project.org mailing list https://stat.ethz.ch/mailman/listinfo/r-help PLEASE do read the posting guide http://www.R-project.org/posting-guide.html and provide commented, minimal, self-contained, reproducible code.
[R] non-linear correlation
Hi all! This is a rather statistical question; Which effect sizes (parametric or not) could I use in order to estimate the amount of non-linear correlation between 2 variables? Is it possible to correct for auto-correlation within the correlated times series? Any suggestions for the appropriate packages/ functions are more than welcome!! Thank you! [[alternative HTML version deleted]] __ R-help@r-project.org mailing list https://stat.ethz.ch/mailman/listinfo/r-help PLEASE do read the posting guide http://www.R-project.org/posting-guide.html and provide commented, minimal, self-contained, reproducible code.
[R] regression with error in predictor
Hi all! I am trying to run a regression where the predictor values are not real data but each is estimated from a different model. So, for each value I have a mean and variance. Which package/function should I use in this case? Thank you! Irene [[alternative HTML version deleted]] __ R-help@r-project.org mailing list https://stat.ethz.ch/mailman/listinfo/r-help PLEASE do read the posting guide http://www.R-project.org/posting-guide.html and provide commented, minimal, self-contained, reproducible code.
[R] mean and variance of ratio
Hi all! I try to estimate a statistic of the form: (x1-x2)/(y1-y2), where x1,x2,y1,y2 represent variable means, so each has an estimate and standard error associated with it. How is it possible to estimate the mean and the variance of this ratio? Thank you! [[alternative HTML version deleted]] __ R-help@r-project.org mailing list https://stat.ethz.ch/mailman/listinfo/r-help PLEASE do read the posting guide http://www.R-project.org/posting-guide.html and provide commented, minimal, self-contained, reproducible code.
[R] autocorrelation by group in mixed model
Hi all! (How) is it possible to fit a mixed model with group specific auto-correlation structure ? For instance, not all my groups display auto-correlation so I would like to use a corMatrix (corAR1) only for the auto-correlated ones. If I construct manually a the corMatrix, is it possible to use it as input somehow? thank you! Irene __ R-help@r-project.org mailing list https://stat.ethz.ch/mailman/listinfo/r-help PLEASE do read the posting guide http://www.R-project.org/posting-guide.html and provide commented, minimal, self-contained, reproducible code.
[R] level significance
Hi all! I am fitting a (mixed) model with a factor (F) and continuous response and predictor: y~F+F:x (How) can I check the significance of the model at each factor level (i.e. the model could be significant only at one of the levels)? Thank you! __ R-help@r-project.org mailing list https://stat.ethz.ch/mailman/listinfo/r-help PLEASE do read the posting guide http://www.R-project.org/posting-guide.html and provide commented, minimal, self-contained, reproducible code.
[R] interaction of continuous terms
Hi all! this is a rather statistical question: is it meaningful to consider an interaction effect between 2 continuous covariates? for example: lm(y~x1+x2+x1:x2) Should one of continuous x1, x2 be transformed to a categorical variable, i.e. be classified into groups? Is it easier to interpret the effect if 1 or both are centered to the mean or z-transformed? Thank you! __ R-help@r-project.org mailing list https://stat.ethz.ch/mailman/listinfo/r-help PLEASE do read the posting guide http://www.R-project.org/posting-guide.html and provide commented, minimal, self-contained, reproducible code.
[R] intercept in multiple regression
Hi all! Is it possible to model a multiple regression in which the response becomes zero when one of the two covariates is zero? lm(y~ x1+x2) and y=0 if x1=0. However, when x1=0, y=x2+1(intercept). Does this mean I cannot have a second covariate and intercept or should I eliminate only the intercept? thank you! __ R-help@r-project.org mailing list https://stat.ethz.ch/mailman/listinfo/r-help PLEASE do read the posting guide http://www.R-project.org/posting-guide.html and provide commented, minimal, self-contained, reproducible code.
Re: [R] interpretation for multiple regression
Hi! Yes, I think that you understood it right and made it clear enough to me too! thank you! :) Από: Daniel Malter [mailto:[EMAIL PROTECTED] Αποστολή: Κυρ 11/11/2007 10:55 μμ Προς: 'Rolf Turner'; Irene Mantzouni Κοιν.: [EMAIL PROTECTED] Θέμα: AW: [R] interpretation for multiple regression Hi, maybe I don't understand your question correctly, but I don't see a reason why there should be a significant interaction term at all. To the question why you regression is linear in B and C: your regressor B is linear (in B), i.e. B= 1B. your regressor C is some quadratic function of B, i.e. C=cB^2 As I see it, your C captures the quadratic form of the relationship between B and A. The relationship between C and A should therefore be linear in the regression because your C already captures the quadratic form. I.e. your C cloaks the significant relationship, because C is itself a quadratic function of B. Does that make any sense? Or did I understand your question incorrectly? Daniel - cuncta stricte discussurus - -Ursprungliche Nachricht- Von: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED] Im Auftrag von Rolf Turner Gesendet: Sunday, November 11, 2007 4:06 PM An: Irene Mantzouni Cc: [EMAIL PROTECTED] Betreff: Re: [R] interpretation for multiple regression On 10/11/2007, at 3:48 AM, Irene Mantzouni wrote: Dear all, probably this is quite clear for most of you but for me it is a headache... I am regressing response A against the continuous covariate B and the relationship is clearly quadratic. When I add a second covariate B, the relationship becomes linear for both B and C. So, I expect that the interaction of B:C should also be significant, which is not the case. How do you interpret this?? Suppose A = a*B + b*C and C = c*B^2. Then A = a*B + b*c+B^2 --- quadratic in B. But the original relationship, predicting A from B and C, is simply linear in B and C, no interaction term. Clear enuff? cheers, Rolf Turner ## Attention:\ This e-mail message is privileged and confid...{{dropped:9}} __ R-help@r-project.org mailing list https://stat.ethz.ch/mailman/listinfo/r-help PLEASE do read the posting guide http://www.R-project.org/posting-guide.html and provide commented, minimal, self-contained, reproducible code. __ R-help@r-project.org mailing list https://stat.ethz.ch/mailman/listinfo/r-help PLEASE do read the posting guide http://www.R-project.org/posting-guide.html and provide commented, minimal, self-contained, reproducible code.
[R] interpretation for multiple regression
Dear all, probably this is quite clear for most of you but for me it is a headache... I am regressing response A against the continuous covariate B and the relationship is clearly quadratic. When I add a second covariate B, the relationship becomes linear for both B and C. So, I expect that the interaction of B:C should also be significant, which is not the case. How do you interpret this?? Thank you! __ R-help@r-project.org mailing list https://stat.ethz.ch/mailman/listinfo/r-help PLEASE do read the posting guide http://www.R-project.org/posting-guide.html and provide commented, minimal, self-contained, reproducible code.
[R] mixed model testing
Is there a formal way to prove the need of a mixed model, apart from e.g. comparing the intervals estimated by lmList fit? For example, should I compare (with AIC ML?) a model with seperately (unpooled) estimated fixed slopes (i.e.using an index for each group) with a model that treats this parameter as a random effect (both models treat the remaining parameters as random)? Thank you! __ R-help@r-project.org mailing list https://stat.ethz.ch/mailman/listinfo/r-help PLEASE do read the posting guide http://www.R-project.org/posting-guide.html and provide commented, minimal, self-contained, reproducible code.
[R] hierarchical mixed model
I would like to fit a 2-level mixed model: yit=a+a[i]+a[it] +(b+b[i]+b[it])*xit+eps[it] However, the variance of the second level components should depend on the group, i.e. sigma for a[it] and b[it] should be [i] specific. I do not know whether this is conceptually right in the mixed model context... In case it stands, how should the formula look like? Also, the data are unbalanced with different number of observations t nested in each i group and I get the warning when trying to fit the model in the traditional way. How much should I worry about this? Thank you! __ R-help@r-project.org mailing list https://stat.ethz.ch/mailman/listinfo/r-help PLEASE do read the posting guide http://www.R-project.org/posting-guide.html and provide commented, minimal, self-contained, reproducible code.
Re: [R] coef se in lme
Thank you very much for the reply (and hopefully I am replying back in the proper way). Do you think the delta method would be an acceptable way to estimate approximate confidence intervals for the resulting group specific coefficients (combining fixed effects and BLUPS)? Regarding the MCMC related approach, how is it possible to save the fixed and the random effects from the MCMC chain? Can this be implemented through nlme library or is there some more clear cut way (I wish I had a strong statistical background and abilities but... :)) to evaluate the empirical distribution of a parameter that is linear combination of these quantities? All the best, Irene Από: [EMAIL PROTECTED] εκ μέρους Douglas Bates Αποστολή: Τετ 17/10/2007 10:04 μμ Προς: Doran, Harold Κοιν.: Irene Mantzouni; [EMAIL PROTECTED]; R-SIG-Mixed-Models Θέμα: Re: [R] coef se in lme On 10/15/07, Doran, Harold [EMAIL PROTECTED] wrote: ?vcov The vcov method returns the estimated variance-covariance matrix of the fixed-effects only. I think Irene's question is about the combination of the fixed-effects parameters and the BLUPs of the random effects that is returned by the coef method applied to an lmer object. (You may recall that you were the person who requested such a method in lme4 like the coef method in nlme :-) On the face of it this quantity should be easy to define and evaluate but in fact it is not easy to do so because these are combinations of model parameters (the fixed effects) and unobserved random variables (the random effects). It gets a bit tricky trying to decide what the variance of this combination would be. I think there is a sensible definition, or at least a computationally reasonable definition, but there are still a few slippery points in the argument. Lately I have taken to referring to the estimates of the random effects, what are sometimes called the BLUPs or Best Linear Unbiased Predictors, as the conditional modes of the random effects. That is, they are the values that maximize the density of the random effects given the observed data and the values of the model parameters. For a linear mixed model the conditional distribution of the random effects is multivariate normal so the conditional modes are also the conditional means. Also, we can evaluate the conditional variance-covariance matrix of the random effects up to a scale factor. The next part is where things get a bit hazy for me but I think it makes sense to consider the joint distribution of the estimator of the fixed-effects parameters and the random effects conditional on the data and, possibly, on the variance components. Conditional on the relative variance-covariance of the random effects (i.e. the matrix that occurs as the penalty term in the penalized least squares representation of the model) the joint distribution of the fixed-effects estimators and the random effects is multivariate normal with mean and variance-covariance matrix determined from the mixed-model equations. This big (p+q by p+q, where p is the dimension of the fixed effects and q is the dimension of the random effects) variance-covariance matrix could be evaluated and, from that, the variance of any linear combination of components. However, I have my doubts about whether it is the most sensible answer to evaluate. Conditioning on the relative variance-covariance matrix of the random effects is cheating, in a way. It would be like saying we have a known variance, $\sigma^2$ when, in fact, we are using an estimate. The fact that we don't know $\sigma^2$ is what gives rise to the t distributions and F distributions in linear models and we are all trained to pay careful attention to the number of degrees of freedom in that estimate and how it affects our ideas of the precision of the estimates of other model parameters. For mixed models, though, many practioners are quite comfortable conditioning on the value of some of the variance components but not others. It could turn out that conditioning on the relative variance-covariance of the random effects is not a big deal but I don't know. I haven't examined it in detail and I don't know of others who have. Another approach entirely is to use Markov chain Monte Carlo to examine the joint distribution of the parameters (in the Bayesian sense) and the random effects. If you save the fixed effects and the random effects from the MCMC chain then you can evaluate the linear combination of interest throughout the chain and get an empirical distribution of the quantities returned by coef. This is probably an unsatisfactory answer for Irene who may have wanted something quick and simple. Unfortunately, I don't think there is a quick, simple answer here. I suggest we move this discussion to the R-SIG-Mixed-Models list which I am cc:ing on this reply. -Original Message- From: [EMAIL PROTECTED] on behalf of Irene Mantzouni Sent: Mon 10/15/2007 3:20 PM To: [EMAIL
[R] coef se in lme
Hi all! How is it possible to estimate standard errors for coef obtained from lme? Is there sth like se.coef() for lmer or what is the anaytical solution? Thank you! __ R-help@r-project.org mailing list https://stat.ethz.ch/mailman/listinfo/r-help PLEASE do read the posting guide http://www.R-project.org/posting-guide.html and provide commented, minimal, self-contained, reproducible code.
[R] weird (?) resid(.) from 2-level lme()
Hi all! I am trying to fit a 2-level hierarchical lme(). I can extract ranef() and coef() without problems for levels=1:2. However, when it comes to resid() at level=2, the resulting list has some unnamed entries (label=NA). I checked with my data (NAs have been omitted) and I found out that it gives 2-3 residuals less in each subgroup. Any ideas what might be wrong? Thank you! Irene __ R-help@r-project.org mailing list https://stat.ethz.ch/mailman/listinfo/r-help PLEASE do read the posting guide http://www.R-project.org/posting-guide.html and provide commented, minimal, self-contained, reproducible code.
Re: [R] continue for loop in case of erros
Thank you all! Yes, try works! __ R-help@r-project.org mailing list https://stat.ethz.ch/mailman/listinfo/r-help PLEASE do read the posting guide http://www.R-project.org/posting-guide.html and provide commented, minimal, self-contained, reproducible code.
[R] constructing a self-starting non-linear model
Dear all, I am trying to define a selfStart function for a non-linear model, which is a log-transformed SSmicmen model with multiplicative errors and so it is required to make them additive: log(y)=log(a)+log(x)-log(1+x/b) Any ideas about how to use the peeling method to derive the initial argument and get the initial values? Thank you for being always there!:) Irene __ R-help@r-project.org mailing list https://stat.ethz.ch/mailman/listinfo/r-help PLEASE do read the posting guide http://www.R-project.org/posting-guide.html and provide commented, minimal, self-contained, reproducible code.
[R] non-linear model parameterization
Dear all, I would like to fit a non-linear model of the form: y=g*x/(a+b*x) with nls(). However this model is somehow overparameterized and I get the error message about singular gradient matrix at initial parameter estimates. What I am interested in is to make inference about parameters b and g, so this has to be taken into account in the model formulation. What options do I have? Also, how is it possible to fit a partially linear model? Thank you!! Irene Mantzouni PhD student DIFRES __ R-help@r-project.org mailing list https://stat.ethz.ch/mailman/listinfo/r-help PLEASE do read the posting guide http://www.R-project.org/posting-guide.html and provide commented, minimal, self-contained, reproducible code.