Re: [R] bootstrapping quantile regression

A possiblie solution might be to use the survey package. You could specify that the data is clustered using the svydesign function, and then speciy the replicate weights using the as.svrepdesign function. And then, it would be possible to use the withReplicates function to bootstrap the clusters

Re: [R] weighted mean by week

Honestly, I wasn't sure what you wanted to do with 'group'. Here it is with the 'group' variable deleted library(data.table) dt=data.table(x[,-1]) dt[,lapply(.SD, function(x)weighted.mean(x,myweight)), keyby='myweek'] -- View this message in context: http://r.789695.n4.nabble.com/weighted

Re: [R] weighted mean by week

If there are many variables, I'd then suggest the data.table package: library(data.table) dt=data.table(x) dt[,lapply(.SD, function(x)weighted.mean(x,myweight)), keyby=c('group', 'myweek')] The '.SD' is an abbreviation for all the variables in the data table (excluding the grouping variables). T

Re: [R] weighted mean by week

The plyr package is very helpful for this: library(plyr) ddply(x ,.(myweek), summarize, m1=weighted.mean(var1,myweight), m2=weighted.mean(var2,myweight)) -- View this message in context: http://r.789695.n4.nabble.com/weighted-mean-by-week-tp4636814p4636816.html Sent from the R help mailing list

Re: [R] Amelia error

I've had a little experience using the package, Amelia. Are you sure that your nominal variables - race, south, etc - are in your ad04 data frame ? David Freedman -- View this message in context: http://r.789695.n4.nabble.com/Amelia-error-tp4568455p4568600.html Sent from the R help ma

Re: [R] Trojan in R 2.13.0?

2.13.0 looks fine with VIPRE david freedman atlanta -- View this message in context: http://r.789695.n4.nabble.com/Trojan-in-R-2-13-0-tp3450084p3450784.html Sent from the R help mailing list archive at Nabble.com. __ R-help@r-project.org mailing

Re: [R] for loop to merge .csvs

you might try merge_recurse(list(DF1,DF2,DF3,DF4,DF17)) in the reshape package -- View this message in context: http://r.789695.n4.nabble.com/for-loop-to-merge-csvs-tp3298549p3298565.html Sent from the R help mailing list archive at Nabble.com. __

Re: [R] Using summaryBy with weighted data

You might use the plyr package to get group-wise weighted means library(plyr) ddply(mydata,~group,summarise, b=mean(weights), c=weighted.mean(response,weights)) hth david freedman -- View this message in context: http://r.789695.n4.nabble.com/Using-summaryBy-with-weighted-data

Re: [R] How to this SAS transport file in R?

I've had good success with the read.xport function in the SASxport (not foreign) package. But couldn't you just use something like mydata<-read.xport('http/demo_c.xpt') The transport file looks like it's one of the NHANES demographic files. -- View this message in context: http://r.789695

Re: [R] data.frame and formula classes of aggregate

Thanks for the information. There was a discussion of different results obtained with the formula and data.frame methods for a paired t-test -- there are many threads, but one is at http://r.789695.n4.nabble.com/Paired-t-tests-td2325956.html#a2326291 david freedman -- View this message in

[R] data.frame and formula classes of aggregate

seems to delete the entire row (missing and non-missing values) if there are any NAs. Wouldn't one expect that the 2 forms (data frame vs formula) of aggregate would give the same result? thanks very much david freedman, atlanta -- View this message in context: http://r.7896

[R] aggregate with missing values, data.frame vs formula

me form deletes gives an NA if there are missing values in the group for the variable with missing values. But, the formula form deletes the entire row (missing and non-missing values) if any of the values are missing. Is this what was intended or the best option ? thanks, david freedman -- View t

Re: [R] Svy function doesn't work nested in user-defined function

I'm not sure that this is the problem, but are you certain that the variable 'con' is in audit ? You check outside the function just really tells you that X and SEX are in audit. hth, david freedman (good to see another CDCer using R) -- View this message in context: htt

Re: [R] percent by subset

how about: d=data.frame(ht=rnorm(20,60,20),sex=sample(0:1,20,rep=T)); d with(d,by(ht,sex,mean)) with(d,by(ht,sex,function(x)mean(x>=60))) hth, david freedman -- View this message in context: http://r.789695.n4.nabble.com/percent-by-subset-tp2123057p2123079.html Sent from the R help mail

Re: [R] binary logistic regression taking account clustering

The bootcov function for models fit using lrm in the rms package might also be an option hth -- View this message in context: http://r.789695.n4.nabble.com/binary-logistic-regression-taking-account-clustering-tp2122255p2122311.html Sent from the R help mailing list archive at Nabble.com. _

Re: [R] help wiht model.matrix -- how to keep missing values?

I'm not sure what to do with model.matrix, but you might look at some of the code for ols in the rms package. The rms package (from frank harrell), has several options for keeping the NAs in the output from regression models so that residuals are correctly aligned. hth david freedman --

Re: [R] get means of elements of 5 matrices in a list

thanks very much for the help - all of these suggestions were much better than what I was doing -- View this message in context: http://r.789695.n4.nabble.com/get-means-of-elements-of-5-matrices-in-a-list-tp2067722p2068329.html Sent from the R help mailing list archive at Nabble.com. __

[R] get means of elements of 5 matrices in a list

y much for any help david freedman ll=list(structure(c(9.7, 17.6, 20.8, 24.1, 33.8, 14.5, 25.7, 29.8, 33.6, 44.8, 21.8, 32.6, 37.5, 40.9, 53.3, 16.7, 26.1, 29.5, 32.7, 42.6, 26.2, 34.3, 37, 39.8, 47.1, 31.9, 40.3, 43.3, 46.2, 54.1 ), .Dim = 5:6), structure(c(9.4, 17.7, 20.7, 24.1, 33.7, 14.5,

Re: [R] Ellipse that Contains 95% of the Observed Data

for a picture of the bagplot, try going to http://www.statmethods.net/graphs/boxplot.html -- View this message in context: http://n4.nabble.com/Ellipse-that-Contains-95-of-the-Observed-Data-tp1694538p1695236.html Sent from the R help mailing list archive at Nabble.com. _

Re: [R] tapply syntax

sorry - I use many abbreviations and I try to remove them before I post questions/answers - 'set' is my abb. for subset david On 3/28/2010 8:27 PM, Jeff Brown [via R] wrote: > What is the function "set()"? Is that a typo? When I type ?set I get > nothing, and when I try to evaluate that code

Re: [R] tapply syntax

how about: d1=data.frame(pat=c(rep('a',3),'b','c',rep('d',2),rep('e',2),'f'),var=c(1,2,3,1,2,2,3,2,4,4)) ds=set(d1,var %in% c(2,3)) with(ds,tapply(var,pat,FUN=length)) hth, David Freedman, CDC, Atlanta -- View this message in context

Re: [R] for() loop

Hi - do you want sum(datjan$V4) OR sum(unique(datjan$V4)) ? there's also a cumsum function that might be useful hth, David Freedman, Atlanta Schmidt Martin wrote: > > Hello > > I'm working with R since a few month and have still many trivial > question

Re: [R] ordering columns in a data frame

ve off the tilde in your orderBy example? hth David Freedman, CDC Atlanta -- View this message in context: http://n4.nabble.com/ordering-columns-in-a-data-frame-tp1587294p1587318.html Sent from the R help mailing list archive at Nabble.com. _

Re: [R] Help with Hmisc, cut2, split and quantile

try as.numeric(read_data$DEC) this should turn it into a numeric variable that you can work with hth David Freedman CDC, Atlanta Guy Green wrote: > > Hi Peter & others, > > Thanks (Peter) - that gets me really close to what I was hoping for. > > The one problem I

Re: [R] How to add a variable to a dataframe whose values are conditional upon the values of an existing variable

there's a recode function in the Hmisc package, but it's difficult (at least for me) to find documentation for it library(Hmisc) week <- c('SAT', 'SUN', 'MON', 'FRI'); recode(week,c('SAT', 'SUN', 'MON', 'FRI'),1:4) HTH -- View this message in context: http://n4.nabble.com/How-to-add-a-variable

Re: [R] Plot of odds ratios obtained from a logistic model

You might want to look at the plot.Predict function in the rms package - it allows you to plot the logits or probablities vs the predictor variable at specified levels of other covariates (if any) in the model. There are many examples in http://cran.r-project.org/web/packages/rms/rms.pdf David

[R] replace a for loop with lapply or relative

(i in 1:3){print(do.call(rbind,by(d[,i],d[,i+3],function(x)(c(min(x),max(x))} Is there a way to replace the ugly for loop in the last line with some type of apply function that would know that my continuous and indicator variable are 3 variables apart in the dataframe? Thanks very much David

Re: [R] tapply for function taking of >1 argument?

also, library(plyr) ddply(d,~grp,function(df) weighted.mean(df$x,df$w)) -- View this message in context: http://n4.nabble.com/tapply-for-function-taking-of-1-argument-tp1460392p1461428.html Sent from the R help mailing list archive at Nabble.com. __

Re: [R] Quartiles and Inter-Quartile Range

Subject: Re: Quartiles and Inter-Quartile Range > > It's looks like you think that type=2 are the 'true' quantiles, but the > default method in R is type=7 > > You might want to look at ?stats::quantile > > hth > david freedman > >

Re: [R] Quartiles and Inter-Quartile Range

w does one calculate the median of 4 values?' david freedman Girish A.R. [via R] wrote: > Interestingly, Hmisc::describe() and summary() seem to be using one > Type, and stats::fivenum() seems to be using another Type. > > > fivenum(cbiomass) > [1] 910.0 1039.0 1088.5 1

Re: [R] Quartiles and Inter-Quartile Range

It's looks like you think that type=2 are the 'true' quantiles, but the default method in R is type=7 You might want to look at ?stats::quantile hth david freedman -- View this message in context: http://n4.nabble.com/Quartiles-and-Inter-Quartile-Range-tp1145817p1213199.html

Re: [R] Mutliple sets of data in one dataset....Need a loop?

You'll probably want to look at the 'by' function d=data.frame(sex=rep(1:2,50),x=rnorm(100)) d$y=d$x+rnorm(100) head(d) cor(d) by(d[,-1],d['sex'],function(df)cor(df)) You might also want to look at the doBy package -- View this message in context: http://n4.nabble.com/Mutliple-sets-of-data-in-

Re: [R] SPLUS Seqtrial vs. R Packages for sequential clinical trials designs

You'll probably want to take a look at the CRAN Task View, http://cran.r-project.org/web/views/ClinicalTrials.html david freedman Paul Miller wrote: > > Hello Everyone, >   > I’m a SAS user who has recently become interested in sequential clinical > trials designs. Iâ

Re: [R] Partial correlations and p-values

you might look at partial.r in the psych package dadrivr wrote: > > I'm trying to write code to calculate partial correlations (along with > p-values). I'm new to R, and I don't know how to do this. I have > searched and come across different functions, but I haven't been able to > get any of

Re: [R] How to duplicate each row in a data.frame?

I *think* this is from from 'StatsRUs' - how about as.data.frame(lapply(df,function(x)rep(x,n))) hth, david freedman pengyu.ut wrote: > > I want to duplicate each line in 'df' 3 times. But I'm confused why > 'z' is a 6 by 4 matrix. Could someb

Re: [R] Weighted descriptives by levels of another variables

In addition to using the survey package (and the svyby function), I've found that many of the 'weighted' functions, such as wtd.mean, work well with the plyr package. For example, wtdmean=function(df)wtd.mean(df$obese,df$sampwt); ddply(mydata, ~cut2(age,c(2,6,12,16)),'w

Re: [R] Weighted descriptives by levels of another variables

In addition to using the survey package (and the svyby function), I've found that many of the 'weighted' functions, such as wtd.mean, work well with the plyr package. For example, wtdmean=function(df)wtd.mean(df$obese,df$sampwt); ddply(mydata, ~cut2(age,c(2,6,12,16)),'w

Re: [R] Using something like the "by" command, but on rows instead of columns

Some variation of the following might be want you want: df=data.frame(sex=sample(1:2,100,replace=T),snp.1=rnorm(100),snp.15=runif(100)) df$snp.1[df$snp.1>1.0]<-NA; #put some missing values into the data x=grep('^snp',names(df)); x #which columns that begin with 'snp' apply(df[,x],2,summary) #or a

Re: [R] Within-group correlation confidence intervals

you should save your 3 variables into a new *dataframe* d<-mydata[,c("iq","education","achievement")] and then the command would be by(d,d$sex,function(df) cor.test(df$educ,df$achiev)) but you could also just use by(mydata,mydata$sex,function(df) cor.test(df$

Re: [R] (no subject)

A better subject for your question might have been helpful. There are many options for hist and truehist in the MASS package, but this might help: x=rnorm(100, mean=5, sd=3000) hist(x, prob=T) x2=density(x) lines(x2$x,x2$y) KABELI MEFANE wrote: > > Dear All >   > I hope you can help me wi

Re: [R] could not find function "Varcov" after upgrade of R?

ry(Hmisc,T) before loading Design? > > Carlos > > ------ > From: "David Freedman" <3.14da...@gmail.com> > Sent: Saturday, September 12, 2009 8:26 AM > To: > Subject: Re: [R] could not find function &q

Re: [R] could not find function "Varcov" after upgrade of R?

nk* it's a problem with the package, rather than R 2.9.2, and I hope the problem will soon be fixed. I was able to use predict.Design with 2.9.2 until I updated the Design package a few days ago. david freedman zhu yao wrote: > > I uses the Design library. > > take this exampl

Re: [R] Calculate weighted mean for each group

After you fix your data frame and if you don't using 2 packages, you might try something like: lib(plyr) #for 'by' processing lib(Hmisc) # for its wtd.mean function d=data.frame(x=c(15,12,3,10,10),g=c(1,1,2,2,3),w=c(2,1,5,2,5)) ; d ddply(d,~g,function(df) wtd.mean(df$x,df$w)) milton ruser wrot

Re: [R] Can I use "mcnemar.test" for 3*3 tables (or is there a bug in the command?)

istic (Stuart, 1955, Agresti, 2002, page 422, also known as Stuart-Maxwell test) is computed." hth, david freedman Tal Galili wrote: > > Hello all, > > I wish to perform a mcnemar test for a 3 by 3 matrix. > By running the slandered R command I am getting a result but I am not

Re: [R] Help With Fleiss Kappa

If you apply the function to a simple dataframe or show your code, you might be able to get more accurate help. I've used the IRR package in the past and haven't noticed any problems (maybe I overlooked them ?) david freedman mehdi ebrahim wrote: > > Hi All, > > I am

Re: [R] Stratified data summaries

You might also want to look at the doBy package - one function is summaryBy: summaryBY(var1 + var2 ~ patient_type, data=d, FUN=summary) david freedman Hayes, Rachel M wrote: > > Hi All, > > > > I'm trying to automate a data summary using summary or describe from

Re: [R] Multiplication of data frame with vector

Would the scale function work for this? Something like new=scale(df, center=T) HTH, david freedman cir p wrote: > > > Dear group: > sorry for my beginners question, but I'm rather new to R and was searching > high and low without success: > > I have a data frame

Re: [R] SAS-like method of recoding variables?

Frank, would you feel comfortable giving us the reference to the NEJM article with the 'missing vs <' error ? I'm sure that things like this happen fairly often, and I'd like to use this example in teaching thanks, david freedman Frank E Harrell Jr wrote:

[R] long format - find age when another variable is first 'high'

000 different children, these functions take some time on my PCs - is there a faster way to do this in R? My code on a small dataset follows. Thanks very much, David Freedman d<-data.frame(id=c(rep(1,3),rep(2,2),3),age=c(5,10,15,4,7,12),ldlc=c(132,120,125,105,142,160)) d$high.ldlc<-ifelse

Re: [R] data summary and some automated t.tests.

You might want to try using a non-parametric test, such as wilcox.test. How about some modification of the following: d=data.frame(grp=rep(1:2,e=5),replicate(10,rnorm(100))); head(d) lapply(d[,-1],function(.column)wilcox.test(.column~grp,data=d)) David Freedman stephen sefick wrote: > &

Re: [R] Help with loops

I'm not quite sure what you want to do, but this might help: d=data.frame(replicate(40, rnorm(20))) d$sample=rep(c('a','b','c','d'),each=5) lib(doBy) summaryBy(.~sample,da=d) David Freedman Amit Patel-7 wrote: > > > Hi > I am try

Re: [R] by-group processing

sorry about the mistake - the -data$Type doesn't work: the '-' sign isn't valid for factors. I *thought* I had checked this before submitting a response ! HufferD wrote: > > On Thursday, May 07, 2009 7:45 PM, David Freedman wrote: > > > ...how about: >

Re: [R] by-group processing

how about: d=data[order(data$ID,-data$Type),] d[!duplicated(d$ID),] Max Webber wrote: > > Given a dataframe like > > > data > ID Type N > 1 45900A 1 > 2 45900B 2 > 3 45900C 3 > 4 45900D 4 > 5 45900E 5 > 6 45900F 6 > 7 45900I 7 > 8

Re: [R] Stepwise logistic regression with significance testing - stepAIC

Didn't a 2008 paper by Austin in J Clin Epidemiol show that bootstrapping was just as bad as backward stepwise regression for finding the true predictors? http://xrl.in/26em Dimitris Rizopoulos-4 wrote: > > Greg Snow wrote: >> There is not a meaningful alternative way since the way you propos

Re: [R] Problems producing a simple plot

you might also look at matplot d=data.frame(x=1:10,y1=sort(rnorm(10)),y2=sort(rnorm(10))) matplot(d[,1],d[,2:3],type='l',lty=1:2) David Freedman Steve Murray-3 wrote: > > > Dear R Users, > > I have a data frame of the nature: > >> head(aggregate_1986)

[R] sequence number for 'long format'

s with the by function, but there must be an easier way. Here's my code - id is id number, age is the age of the person, and seq is the sequence variable that I've created. Thanks very much for the help. david freedman, atlanta ds=data.frame(list(id = c(1L, 1L, 1L, 1L, 8L, 8L, 16L, 16L

[R] piecewise linear regression

1/se, subset=year>=1988, da=d); points(d$year,predict(m,dafr(year=d$year)),type='l',lwd=2,col='red') thanks very much David Freedman -- View this message in context: http://www.nabble.com/piecewise-linear-regression-tp22388118p22388118.html Sent from the R help mailin

Re: [R] calculating standard deviation and variance

The gplots package might help. Try: library(gplots) data(state) plotmeans(state.area ~ state.region) hth, David Freedman johnhj wrote: > > Hiii, > > I have some problems to plot "standard deviation and variance" from a > texfile. > > Ich have the following c

Re: [R] beginner's question: group of regressors by name vector?

The predictors and outcomes in lm can be matrices, so you could use something like the following: x.mat=cbind(x1=rnorm(20),x2=rnorm(20)) y.mat=cbind(y1=rnorm(20),y2=rnorm(20)) lm(y.mat~x.mat) David Freedman ivowel wrote: > > dear r-experts: there is probably a very easy way to do it,

Re: [R] R equivalent of SAS Cochran-Mantel-Haenszel tests?

the linear-by-linear association test for ordered data. David Freedman Michael Friendly wrote: > > In SAS, for a two-way (or 3-way, stratified) table, the CMH option in > SAS PROC FREQ gives > 3 tests that take ordinality of the factors into account, for both > variables, just the

Re: [R] Roc curves confidence intervals

Would you be interested in the cross-validation that's on pp 3 and 4 of the ROCR package PDF? plot(perf,lwd=3,avg="vertical",spread.estimate="boxplot",add=TRUE) there are various options for the 'spread.estimate' David Freedman marc bernard-2 wrote: >

Re: [R] overlay plot question

You could change the second 'plot' to 'points' David Freedman David Kaplan-2 wrote: > > Greetings all, > > I have two logistic plots coming from two calls to plogis. The code is > > .x <- seq(-7.6, 7.6, length=100) > plot(.x, plogis(.x, l

Re: [R] lapply and aggregate function

You might want to look at the doBy package For (1), you could use summaryBy(value~Light+Feed,data=myD, FUN=mean) and for (2), the transformBy function would be helpful David Freedman Patrick Hausmann wrote: > > Dear list, > > I have two things I am struggling... > > # Fi

Re: [R] problem with subsetting in compating one column with a vector

Do you really want to use '==' ? How about '%in%', as in findings1<-subset(findings,SUBJECTSID %in% ==SUBJECTS1$SUBJECTSID, select=c(SUBJECTSID,ORGNUMRES)) David Freedman dvkirankumar wrote: > > Hi all, > I got one problem with"subset()"fun

Re: [R] how to prevent duplications of data within a loop

or how about lm(myData$response~as.matrix(myData[2:4])) hth, david Juliet Hannah wrote: > > Hi All, > > I had posted a question on a similar topic, but I think it was not > focused. I am posting a modification that I think better accomplishes > this. > I hope this is ok, and I apologize if it

Re: [R] Is there any function can be used to compare two probit models made from same data?

Hi - wouldn't it be possible to bootstrap the difference between the fit of the 2 models? For example, if one had a *linear* regression problem, the following script could be used (although I'm sure that it could be improved): library(MASS); library(boot) #create intercorrelated data Sigma <- ma

Re: [R] should I use rbind in my example?

(2,1,100)) library(doBy) summaryBy(wt+ht~sex,da=d,FUN=c(mean,sd)) David Freedman SNN wrote: > > > Hi, > > This is just for print out so it looks nice. what I have is a loop that > loops over my variables and calculates the mean and the sd for these > variables. Then I need t

Re: [R] conditional weighted quintiles

ero. > > Thanks in advance, > -- > ozan bakis >--__--__-- > > __ > 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-g

Re: [R] tapply within a data.frame: a simpler alternative?

You might take a look at the transformBy function in the doBy package For example, new.df=transformBy(~group,data=my.df, new=y/max(y)) David Freedman baptiste auguie-2 wrote: > > Dear list, > > I have a data.frame with x, y values and a 3-level factor "group", >

Re: [R] partial correlation

not sure which library 'pcor' is in, but why don't you just use the ranks of the variables and then perform the correlation on the ranks: x<-sample(1:10,10,rep=T) y<-x+ sample(1:10,10,rep=T) cor(x,y) cor(rank(x),rank(y)) HTH david freedman Jürg Brendan Logue wrote: &g

Re: [R] unlist & dataframes

try newdata=do.call(rbind,l) David Freedman, Atlanta Naira wrote: > > Dear all, > > I would like to know whether it is possible to unlist elements and keep > the original format of the data. > To make it more clear, let me give an exemple: > I have a list l of data

Re: [R] ggplot2 help

You might want to take a look at http://had.co.nz/ggplot2/ There are many examples of how to use this function. And, there's a book coming out soon. hope this helps david freedman Edna Bell wrote: > > Hi yet again! > > Thank you for being patient with me. > >

[R] fill in area between 2 lines with a color

rs)') lines(age,y.low,col='grey') lines(age,y.high,col='grey') Is it possible to fill the area between the 2 lines filled with, for example, 'grey30' ? thanks very much in advance, David Freedman - David Freedman Atlanta -- View this message in context: htt