Re: [R] svymeans question
On 26/8/08 9:40 AM, Doran, Harold wrote: Computing the mean of the item by itself with svymeans agrees with the sample mean mean(lower_dat$W524787, na.rm=T) [1] 0.8555471 Compare this to the value in the row 9 up from the bottom to see it is different. You might be omitting more cases due to missing values than you expect. Does the following calculation give you the same results as in rr1? mean( lower_dat$W524787[ apply( lower_dat[lset], 1, function(x) !any(is.na(x)) ) ] ) James -- James Reilly Department of Statistics, University of Auckland Private Bag 92019, Auckland, New Zealand __ 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] Problems using rfImpute
The values NA and "NA" are different. The first is treated as missing; the second is not. For example, > table(factor(c(NA,"0","1","NA","NA"))) 0 1 NA 1 1 2 I suspect you have "NA" where you want NA, and this is causing your problem. James -- James Reilly Department of Statistics, University of Auckland Private Bag 92019, Auckland, New Zealand On 6/5/08 1:04 AM, Birgit Lemcke wrote: Hello R-user! I am running R 2.7.0 on a Power Book (Tiger). (I am still R and statistics beginner) I tried rfImpute (randomForest) and as far as I understood should it replace NA`s using a proximity matrix: > set.seed(10) > Subset5Imputed<-rfImpute(Sex~., data=Subset5) ntree OOB 1 2 300: 11.78% 12.36% 11.21% ntree OOB 1 2 300: 12.07% 12.64% 11.49% ntree OOB 1 2 300: 11.49% 11.21% 11.78% ntree OOB 1 2 300: 12.50% 12.93% 12.07% ntree OOB 1 2 300: 12.07% 12.36% 11.78% > str(Subset5Imputed) 'data.frame':696 obs. of 24 variables: $ Sex: Factor w/ 2 levels "0","1": 2 2 2 2 2 2 2 2 2 2 ... $ InfSpath_caducuous : Factor w/ 3 levels "0","1","NA": 1 1 1 1 1 1 1 1 1 1 ... $ InfType_sparsely_paniculate: Factor w/ 3 levels "0","1","NA": 1 1 1 3 1 1 1 1 1 1 ... But there are still NA`s in the data frame. Sorry if this reason is only ma stupididty and thanks for answering in advance. B. Birgit Lemcke Institut für Systematische Botanik Zollikerstrasse 107 CH-8008 Zürich Switzerland Ph: +41 (0)44 634 8351 [EMAIL PROTECTED] 175 Jahre UZH «staunen.erleben.begreifen. Naturwissenschaft zum Anfassen.» MNF-Jubiläumsevent für gross und klein. 19. April 2008, 10.00 Uhr bis 02.00 Uhr Campus Irchel, Winterthurerstrasse 190, 8057 Zürich Weitere Informationen http://www.175jahre.uzh.ch/naturwissenschaft __ 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] Imputation Packages
On 3/3/08 6:42 AM, ArunPrasad wrote: > Hi everyone, > I am looking for a package in R which can help me in using the > imputation technique to find the missing values for my regression analysis. > Any help would be appreciated. > > Cheers > Arun Have a look at the "Missing data" sections of the Multivariate and SocialSciences task views on CRAN: http://cran.r-project.org/web/views/ James -- James Reilly Department of Statistics, University of Auckland Private Bag 92019, Auckland, New Zealand __ 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] Studdy Missing Data, differentiate between a percent with in the valid answers and with in the different missing answers
On 3/3/08 8:21 PM, Ericka Lundström wrote:
> I'm trying to emigrate from SPSS to R, thou I have some problems whit
> getting R to distinguish between the different kind of missing.
...
> Is there a smart way in R to differentiate between missing and valid
> and at the same time treat both the categories within missing and
> valid as answers (like SPSS did above)
The Hmisc package has some support for special missing values, for
instance when reading in SAS datasets using sas.get. I don't believe
spss.get offers the same facility, though.
You can define special missing values for a variable manually, which
might seem a bit involved, but this could easily be automated. For your
example, try:
special <- dataFrame$TWO %in% c("?","X")
attr(dataFrame$TWO, "special.miss") <-
list(codes=as.character(dataFrame$TWO[special]),
obs=(1:length(dataFrame$TWO))[special])
class(dataFrame$TWO) <- c("factor", "special.miss")
is.na(dataFrame$TWO) <- special
# Then describe gives new percentages
describe(dataFrame$TWO)
dataFrame$TWO
n missing ? X unique
3 4 2 2 2
No (2, 67%), yes (1, 33%)
HTH,
James
--
James Reilly
Department of Statistics, University of Auckland
Private Bag 92019, Auckland, New Zealand
__
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] things that are difficult/impossible to do in SAS or SPSS but simple in R
> Max Kuhn wrote: >> Factors have huge benefits over character data in SAS. For a series >> regulatory filings, I had miles of SAS code to compute KxK tables >> where all the cells must show up. For example, if one of the levels of >> one of the variables was never observed, the corresponding row or >> column would not show up in proc freq. The basic way around this was >> to get all possible combinations of the variables and assign each cell >> to have a row count of 0.0001. Then you would merge this data with >> the real counts. The missing row/columns would show up since they had >> data, but it was below the printing threshold of proc freq. Hoepfully, >> they have added a feature to do this. >> On 18/1/08 4:44 AM, Peter Dalgaard wrote: > I could have sworn that this was a fluke and that it would work if you > put a user-defined format on the classification variable, but no go > I can't find anything that does this, neither in PROC FREQ nor PROC > TABULATE. I believe the CLASSDATA option in PROC TABULATE lets you specify which values will show up in the table, including unobserved values. http://support.sas.com/onlinedoc/913/getDoc/en/proc.hlp/a002473736.htm#a003069171 I'm not aware of any way to do this in PROC FREQ, though. -- James Reilly Department of Statistics, University of Auckland Private Bag 92019, Auckland, New Zealand __ 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] 4 questions regarding hypothesis testing, survey package, ts on samples, plotting
On 19/12/07 8:39 PM, eugen pircalabelu wrote: > 2. How can i find the median of a variable in survey package? > a<-c(1:10) > b<-sample(1:20, 10, replace=T) > b1<-sample(0:1, 10, replace=T) > c<-data.frame(a,b, b1) > library(survey) > design<-svydesign(id=~1, data=c) > svymean(~b, design) > > svymean gives me the mean, but what function gives me the svymedian, and what > function gives me the svyproportion for b1? > I want to know the median for a metric variable and the proportion for a > binomial variable ? The svyquantile function will give you the median, e.g. svyquantile(~b, design, 0.5, interval.type="score") Since a proportion is just the mean of a 0/1 variable, you can use svymean for that, e.g. svymean(~b1, design) HTH, James -- James Reilly Department of Statistics, University of Auckland Private Bag 92019, Auckland, New Zealand __ 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] spss, string factors, selecting
It does sound like there could be a problem with the merging process. I
have two questions about your merge command:
chaffmerge2<-merge(chaff, chafffat, by.x=c("RINGNO", "FAT",
"FATMTD"), by.y=c("RINGNO", "FAT", "FATMTD"), all=T)
1. What is the reason for matching on "FAT" and "FATMTD"? From your
description of the data, I assume that "RINGNO" is the individual
identifier. I'd have thought matching on that alone would be appropriate.
2. What happens if you omit the "all=T" argument? In particular, how
does the size of the merged dataset compare to the inputs?
--
James Reilly
Department of Statistics, University of Auckland
Private Bag 92019, Auckland, New Zealand
On 27/11/07 2:31 PM, Katherine Jones wrote:
> Hi,
>
> This is probably a case where someone has to see what is happening on
> my computer and it is complicated by my data being from SPSS (not my
> choice). It is quite hard to give my data, because it is such a large
> dataset. I have analysed 9 other datasets that work fine, but this
> particular dataset was inputted wrong so requires merging of two
> datasets. This may be the problem.
>
> Example of data:-
> File 1.
> [1] Individual [2] Habitat type [3] Weight
> File 2.
> [1] Individual [2] Fat [3] Fat method.
>
> I merge the two files to create:-
> [1] Individual [2] Habitat type [3] Weight [4] Fat [5] Fat method
>
> My merging appears to work in the sense that I can plot Weight versus
> Fat and I get data, but if I ask to see the data file I see a sea of
> "NAs". So I'm not sure how there can be data there to plot, see
> levels for and create tables for but I can't see it as a dataframe. I
> do get the plot I want.
>
> Fat method contains either blank cells, " B" or " E".
>
> I wish to select all the rows in columns 1-4 which contain an " E" in
> Fat method.
>
> e.g.
> 120, 3, 20.2, 4, E
> 121, 4, 20.0, 5, B
> 132, 3, 21.2, 4,
>
> I want to select only the row containing " E", so I can plot Fat vs
> Habitat and Weight vs. Fat.
>
> I have been doing this by using
>
> selectE<-Data[Fatmethod==" E",].
>
> However, this does not work. It removes all of my data in the other
> columns to "NA" and I am left only with fatmethod and fat scores.
>
> It is odd it works with other datasets but not this one. Although
> with my other datasets when I ask to select " E", I can still see "
> B" using levels(Fat method) but there is no data there, so my plots
> are correct.
>
> Sorry this is long. I'm having difficulty explaining it.
>
> Katherine
>
>
> On 26-Nov-07, at 5:09 PM, jim holtman wrote:
>
>> That should give you back a subset of 'data' (with all its columns),
>> for those with " E" in 'column'. Can you show an example of your data
>> and what the desired output would be. The posting guide asks "provide
>> commented, minimal, self-contained, reproducible code" so we don't
>> have to speculate on what you want.
>>
>> On Nov 26, 2007 5:04 PM, Katherine Jones
>> <[EMAIL PROTECTED]> wrote:
>>> This sort of works. It does select the E data, but unfortunately
>>> it doesn't
>>> select the data from the other columns; I want to select data
>>> across about 5
>>> columns by the factor " E" in one of the columns. It should be
>>> easy, but for
>>> some reason it is not working. The spaces being added don't help.
>>>
>>> It seems to work on my non-merged data files, although the merged
>>> file
>>> contains all the data I need.
>>>
>>> Thanks for the subset command though. Hadn't thought of using that.
>>>
>>>
>>>
>>> On 26-Nov-07, at 4:46 PM, jim holtman wrote:
>>> ?subset
>>>
>>>
>>> subset(data, column == " E")
>>>
>>
>>
>> --
>> Jim Holtman
>> Cincinnati, OH
>> +1 513 646 9390
>>
>> What is the problem you are trying to solve?
>
> __
> 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.
Re: [R] question regarding an error message from survey package
Your calibration model has one linear predictor, so it only has two
parameters, but you have specified three population totals. A slightly
different setup should work; try again with:
st <- factor(1:3, levels=c(3, 1:2))
pop<-c('(Intercept)'=100, st1=10, st2=20)
James Reilly
Department of Statistics, University of Auckland
Private Bag 92019, Auckland, New Zealand
On 25/11/07 6:47 AM, eugen pircalabelu wrote:
> Hi R-users!
> Can anyone tell me what this error message refers to exactly
>
>> a<-c('m','f','m')
>> b<-c(1:3)
>> st<-c(1:3)
>> data.frame(a,b,st)
> a b st
> 1 m 1 1
> 2 f 2 2
> 3 m 3 3
>> data.frame(a,b,st)->l
>> p<-svydesign(ids=~1, data=l)
>> pop<-c('(intercept)'=100, st1=10,st2=20)
>> calibrate(p,~st, pop)
> Error in regcalibrate.survey.design2(design, formula, population,
> aggregate.stage = aggregate.stage, :
> Population and sample totals are not the same length.
>
> Thank you and have a great day!
>
>
> -
>
> [[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-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] survey weights in sample with replacement
On 31/10/07 4:03 PM, Azam, Mehtabul wrote: >I am trying to draw a random sample from an household survey with > sample weight. Is there any function in R or Splus which allows this. I'm not sure if this what you're after, but R's pps package lets you sample with probability proportional to size. Hope this helps, James -- James Reilly Department of Statistics, University of Auckland Private Bag 92019, Auckland, New Zealand __ 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] cross tabulation
On 14/10/07 5:39 PM, Louis Martin wrote:
> I am looking for an efficient and fast way for doing this:
>
> for (v in 1:dw) {ok <-list1[v,]
>
> for (z in 1:5) {kk = ok[z] ;
> for (o in 1:t) {if (kk== list[o]) n<- n+1}}
> show(n) }
>
> list1 : a data.frame with nrow = dw ; ncol = z;
> list2 : a data.frame with nrow = 1 ; ncol = t
> with t > z
Is the following close to what you want?
cumsum(apply(list1[,1:5], 1, function(x) sum(x %in% list2)))
Here list2 is a vector, not a data frame. Duplicates in list2 are not
counted twice, unlike in your code, but you could run it again just on
the duplicates (and again on the triplicates, etc) if this is needed.
Hope this helps,
James
--
James Reilly
Department of Statistics, University of Auckland
Private Bag 92019, Auckland, New Zealand
__
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] Mac GUI and .Renviron
On 7/10/07 3:49 AM, Rob J Goedman wrote: > I do not think you can, without some further steps, run multiple > R.apps at the same time. Let me know if that is critical for you. If you copy R.app, you can run each copy concurrently. James -- James Reilly Department of Statistics, University of Auckland Private Bag 92019, Auckland, New Zealand __ 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] Re-ordering factors
Using reorder.factor from the stats package seems to work:
educ$ed <- reorder(educ$Education, sort(rep(1:4,5)))
levels(educ$Education)
[1] "CompletedHS" "IncompleteHS" "Uni1-3" "Uni4+"
levels(educ$ed)
[1] "IncompleteHS" "CompletedHS" "Uni1-3" "Uni4+"
xtabs(Count ~ ed + Age_Group, data=educ)
Age_Group
ed 25-34 35-44 45-54 55-64 >64
IncompleteHS 5416 5030 5777 7606 13746
CompletedHS 16431 1855 9435 8795 7558
Uni1-38555 5576 3124 2524 2503
Uni4+ 9771 7596 3904 3109 2483
James
On 7/10/07 4:52 PM, [EMAIL PROTECTED] wrote:
> A small example before I begin my query:
>
>> educ <- read.table(efile, header=TRUE)
>> educ
> Education Age_Group Count
> 1 IncompleteHS 25-34 5416
> 2 IncompleteHS 35-44 5030
> 3 IncompleteHS 45-54 5777
> 4 IncompleteHS 55-64 7606
> 5 IncompleteHS >64 13746
> 6 CompletedHS 25-34 16431
> 7 CompletedHS 35-44 1855
> 8 CompletedHS 45-54 9435
> 9 CompletedHS 55-64 8795
> 10 CompletedHS >64 7558
> 11 Uni1-3 25-34 8555
> 12 Uni1-3 35-44 5576
> 13 Uni1-3 45-54 3124
> 14 Uni1-3 55-64 2524
> 15 Uni1-3 >64 2503
> 16Uni4+ 25-34 9771
> 17Uni4+ 35-44 7596
> 18Uni4+ 45-54 3904
> 19Uni4+ 55-64 3109
> 20Uni4+ >64 2483
>> xtabs(Count ~ Education + Age_Group, data=educ)
> Age_Group
> Education>64 25-34 35-44 45-54 55-64
> CompletedHS 7558 16431 1855 9435 8795
> IncompleteHS 13746 5416 5030 5777 7606
> Uni1-32503 8555 5576 3124 2524
> Uni4+ 2483 9771 7596 3904 3109
>
> Naturally I would prefer the factor levels in their natural ordering in
> the table. I would like to re-order the levels of the factors to achieve
> this.
>
> I have tried reorder() in the gdata package:
>
>> ed <- reorder(Education,neworder= c("IncompleteHS","CompletedHS",
> + "Uni1-3","Uni4+"))
>> agrp <- reorder(Age_Group,neworder=
> + c("25-34","35-44","45-54","55-64",">64"))
>> xtabs(Count ~ ed + agrp)
> agrp
> ed 25-34 35-44 45-54 55-64 >64
> CompletedHS 16431 1855 9435 8795 7558
> IncompleteHS 5416 5030 5777 7606 13746
> Uni1-38555 5576 3124 2524 2503
> Uni4+ 9771 7596 3904 3109 2483
>
> which works on one factor but not the other.
>
> I have fooled a bit with reorder.factor() but I can't seem to figure out
> how to drive it!
>
> Cheers, Murray Jorgensen
--
James Reilly
Department of Statistics, University of Auckland
Private Bag 92019, Auckland, New Zealand
__
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] Ambiguities in vector
If I understand you right, you have several multiple response variables
(with the responses encoded in numeric strings) and you want to see
whether these are associated with sex. To tabulate the data, I would
convert your variables into collections of dummy variables using
regexpr(), then use table(). You can use a modified chi-squared test
with a Rao-Scott correction on the resulting tables; see Thomas and
Decady (2004). Bootstrapping is another possible approach.
@article{,
Author = {Thomas, D. Roland and Decady, Yves J.},
Journal = {International Journal of Testing},
Number = {1},
Pages = {43 - 59},
Title = {Testing for Association Using Multiple Response Survey Data:
Approximate Procedures Based on the Rao-Scott Approach.},
Volume = {4},
Year = {2004},
Url=http://search.ebscohost.com/login.aspx?direct=true&db=pbh&AN=13663214&site=ehost-live
}
Hope this helps,
James
--
James Reilly
Department of Statistics, University of Auckland
Private Bag 92019, Auckland, New Zealand
On 21/9/07 7:14 AM, Birgit Lemcke wrote:
> First thanks for your answer.
> Now I try to explain better:
>
> I have species in the rows and morphological attributes in the
> columns coded by numbers (qualitative variables; nominal and ordinal).
> In one table for the male plants of every species and in the other
> table for the female plants of every species. The variables contain
> every possible occurrence in this species and this gender.
> I would like to compare every variable between male and female plants
> for example using a ChiSquare Test.
> The Null-hypothesis could be: Variable male is equal to variable Female.
>
> The question behind all is, if male and female plants in this species
> are significantly different and which attributes are responsible for
> this difference.
>
> I really hope that this is better understandable. If not please ask.
>
> Thanks a million in advance.
>
>
> Greetings
>
> Birgit
__
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] Multiple R sessions, Mac version
On 14/9/07 3:00 AM, David Afshartous wrote: > I've just switched to running R 2.5.1 on a Mac 0S X 10.4.1 platform. I > can't seem to find how to run simultaneous R sessions. Didn't see anything > in the archives on this or under the R file menu. I've switched to Mac OS X recently too (10.4.10). While running simultaneous R sessions under X11 or from the terminal works for me out of the box, the only way I've found to run simultaneous Aqua sessions is to make extra copies of R.app and run those. James -- James Reilly Department of Statistics, University of Auckland Private Bag 92019, Auckland, New Zealand __ 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] Missing data
The mice package might also be useful, especially the md.pattern function: http://finzi.psych.upenn.edu/R/library/mice/html/md.pattern.html James -- James Reilly Department of Statistics, University of Auckland Private Bag 92019, Auckland, New Zealand On 12/9/07 1:33 PM, Bill Pikounis wrote: > David, > Frank Harrell's pair of packages Hmisc and Design has some functions > for tabulating, visualizing, and accounting for missing data. I > recommend them as one avenue to investigate. Frank's companion book > "Regression Modeling Strategies" covers their use in-depth. > > Hope that helps, > Bill > > ___ > Bill Pikounis > Statistician > > On 9/11/07, David Kaplan <[EMAIL PROTECTED]> wrote: >> Hi all, >> >> I'm looking for a contributed package that can provide a detailed >> account of missing data patterns and perhaps also provide imputation >> procedures, such as mean imputation or hot deck imputation and the like. >> Is there anything out there? >> >> Thanks in advance, >> >> David >> __ 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.
