Re: [R] R2WinBUGS results not different with different runs
Gregor Gorjanc wrote: toby909 at gmail.com writes: Is this a specialty with R2WinBUGS? Does it have something to do with the seed value? Isnt the seed value reset everytime I restart winbugs? I always have the same seed if I start WinBUGS multiple times. So you get exactly the same chain, numerically, when rerunning the same model, with the same number of iterations, everything the same.? Wouldnt that be problematic if every researcher in the world who uses winbugs uses the same sequence of random numbers? R's random numbers are different each time, because the seed is linked to the clock in your PC. T Gregor __ R-help@stat.math.ethz.ch mailing list https://stat.ethz.ch/mailman/listinfo/r-help PLEASE do read the posting guide http://www.R-project.org/posting-guide.html and provide commented, minimal, self-contained, reproducible code. __ R-help@stat.math.ethz.ch mailing list https://stat.ethz.ch/mailman/listinfo/r-help PLEASE do read the posting guide http://www.R-project.org/posting-guide.html and provide commented, minimal, self-contained, reproducible code.
Re: [R] R2WinBUGS results not different with different runs
toby909 at gmail.com writes: Gregor Gorjanc wrote: toby909 at gmail.com writes: Is this a specialty with R2WinBUGS? Does it have something to do with the seed value? Isnt the seed value reset everytime I restart winbugs? I always have the same seed if I start WinBUGS multiple times. So you get exactly the same chain, numerically, when rerunning the same model, with the same number of iterations, everything the same.? I just tried now with Gelman's schools example from bugs() help page. I runned the same job twice, with exactly the same initial values. Notes that WinBUGS uses always the same starting seed. I got this schools.sim$last.values [[1]] [[1]]$theta [1] 23.700 10.060 12.760 13.090 1.693 14.390 7.599 3.961 [[1]]$mu.theta [1] 25.03 [[1]]$sigma.theta [1] 20.08 schools.sim2$last.values [[1]] [[1]]$theta [1] 23.700 10.060 12.760 13.090 1.693 14.390 7.599 3.961 [[1]]$mu.theta [1] 25.03 [[1]]$sigma.theta [1] 20.08 Wouldnt that be problematic if every researcher in the world who uses winbugs uses the same sequence of random numbers? R's random numbers are different each time, because the seed is linked to the clock in your PC. You can reset the seed if you want. Having the same seed is also nice for repeatability. Gregor __ R-help@stat.math.ethz.ch mailing list https://stat.ethz.ch/mailman/listinfo/r-help PLEASE do read the posting guide http://www.R-project.org/posting-guide.html and provide commented, minimal, self-contained, reproducible code.
[R] R2WinBUGS results not different with different runs
Hi All I dont know if anyone else has noticed the same thing, but with 2 subsequent runs of the same syntax, I am getting exactly the same results. I was expecting that results differ slighlty, say in the 4th or 5th decimal place. Is this a specialty with R2WinBUGS? Does it have something to do with the seed value? Isnt the seed value reset everytime I restart winbugs? Thanks Toby __ R-help@stat.math.ethz.ch mailing list https://stat.ethz.ch/mailman/listinfo/r-help PLEASE do read the posting guide http://www.R-project.org/posting-guide.html and provide commented, minimal, self-contained, reproducible code.
Re: [R] R2WinBUGS results not different with different runs
toby909 at gmail.com writes: Is this a specialty with R2WinBUGS? Does it have something to do with the seed value? Isnt the seed value reset everytime I restart winbugs? I always have the same seed if I start WinBUGS multiple times. Gregor __ R-help@stat.math.ethz.ch mailing list https://stat.ethz.ch/mailman/listinfo/r-help PLEASE do read the posting guide http://www.R-project.org/posting-guide.html and provide commented, minimal, self-contained, reproducible code.
Re: [R] R2WinBUGS more updates after model did not converge
[EMAIL PROTECTED] wrote: After running a model for a while and seeing that it did not converge yet, how can I continue to run, ie not starting anew, the model? I know if I manually/interactively use winbugs, this is possible anytime, but how can I do this in r2winbugs, so that my existing sim$sims.array and other stuff in the object that bugs() returns gets extended? This is not possible with R2WinBUGS. But you can use BRugs, which is much more flexible, has an improved interface and uses OpenBUGS. It's CRAN version is based on OpenBUGS 2.x.y, but we (i.e. currently mainly Insightful people and myself) are on the way to release a new version based on OpenBUGS 3.x.y. Uwe Ligges Thanks Toby __ R-help@stat.math.ethz.ch mailing list https://stat.ethz.ch/mailman/listinfo/r-help PLEASE do read the posting guide http://www.R-project.org/posting-guide.html and provide commented, minimal, self-contained, reproducible code. __ R-help@stat.math.ethz.ch mailing list https://stat.ethz.ch/mailman/listinfo/r-help PLEASE do read the posting guide http://www.R-project.org/posting-guide.html and provide commented, minimal, self-contained, reproducible code.
[R] R2WinBUGS more updates after model did not converge
After running a model for a while and seeing that it did not converge yet, how can I continue to run, ie not starting anew, the model? I know if I manually/interactively use winbugs, this is possible anytime, but how can I do this in r2winbugs, so that my existing sim$sims.array and other stuff in the object that bugs() returns gets extended? Thanks Toby __ R-help@stat.math.ethz.ch mailing list https://stat.ethz.ch/mailman/listinfo/r-help PLEASE do read the posting guide http://www.R-project.org/posting-guide.html and provide commented, minimal, self-contained, reproducible code.
Re: [R] R2WinBUGS awkward to use
[EMAIL PROTECTED] wrote:
Hi All
Does anyone know if I can avoid to use the write.model() function below? I
dont
want to do this. Can't bugs() do that automatically for me just by specifying
the 4th argument 'model'? Just I like I am also using the 'inits' object!
If I use 'model' in the same way as I use 'inits' I am getting the error:
sim - bugs(data, inits, parameters, model, n.chains=1, n.iter=5000,
bugs.directory=c:/Program Files/WinBUGS14, working.directory=NULL,
clearWD=TRUE, DIC=0)
Error in file.exists(c(...)) : invalid 'file' argument
The idea is that people in general write their model files more or less
while interactively using WinBUGS / OpenBUGS and later on start
automizing with R2WinBUGS or BRugs. Then the file already exists.
Of course, we can add dome fancy stuff that applies write.model if an
expression is given in the argument model.
I'll put it on the ToDo list.
Best,
Uwe Ligges
Thanks Toby
data - list(.)
model - function() {
[omitted]
}
write.model(model, cwk.txt)
inits - function() {.}
parameters - c(b, nu, S1, S2)
sim - bugs(data, inits, parameters, cwk.txt, n.chains=1, n.iter=5000,
bugs.directory=c:/Program Files/WinBUGS14, working.directory=NULL,
clearWD=TRUE, DIC=0)
print(sim)
plot(sim)
__
R-help@stat.math.ethz.ch mailing list
https://stat.ethz.ch/mailman/listinfo/r-help
PLEASE do read the posting guide http://www.R-project.org/posting-guide.html
and provide commented, minimal, self-contained, reproducible code.
__
R-help@stat.math.ethz.ch mailing list
https://stat.ethz.ch/mailman/listinfo/r-help
PLEASE do read the posting guide http://www.R-project.org/posting-guide.html
and provide commented, minimal, self-contained, reproducible code.
[R] R2WinBUGS awkward to use
Hi All
Does anyone know if I can avoid to use the write.model() function below? I dont
want to do this. Can't bugs() do that automatically for me just by specifying
the 4th argument 'model'? Just I like I am also using the 'inits' object!
If I use 'model' in the same way as I use 'inits' I am getting the error:
sim - bugs(data, inits, parameters, model, n.chains=1, n.iter=5000,
bugs.directory=c:/Program Files/WinBUGS14, working.directory=NULL,
clearWD=TRUE, DIC=0)
Error in file.exists(c(...)) : invalid 'file' argument
Thanks Toby
data - list(.)
model - function() {
[omitted]
}
write.model(model, cwk.txt)
inits - function() {.}
parameters - c(b, nu, S1, S2)
sim - bugs(data, inits, parameters, cwk.txt, n.chains=1, n.iter=5000,
bugs.directory=c:/Program Files/WinBUGS14, working.directory=NULL,
clearWD=TRUE, DIC=0)
print(sim)
plot(sim)
__
R-help@stat.math.ethz.ch mailing list
https://stat.ethz.ch/mailman/listinfo/r-help
PLEASE do read the posting guide http://www.R-project.org/posting-guide.html
and provide commented, minimal, self-contained, reproducible code.
Re: [R] R2WinBugs and Compare DIC versus BIC or AIC
Thanks for your answer !
But If I have understood correctly, the procedure is correct and this only
means that I have already generated the initial values
From R. I define a function to generate the initial values as:
inits. - function(){list(sdesp=runif(1),sdhet=runif(1),beta0=runif(1),
beta1=runif(1), b=rnorm(n), phi=rnorm(n))}
then It is not necessary to generate initial values, that why gen.inits() is
greyed out
, is it ok?
Here is what Winbugs show
==
display(log)
check(D:/model.txt)
model is syntactically correct
data(D://data.txt)
data loaded
compile(3)
model compiled
inits(1,D:/inits1.txt)
chain initialized but other chain(s) contain uninitialized variables
inits(2,D:/inits2.txt)
chain initialized but other chain(s) contain uninitialized variables
inits(3,D:/inits3.txt)
model is initialized
gen.inits()
command #Bugs:gen.inits cannot be executed (is greyed out)
thin.updater(1)
update(5000)
==
Thank again in advance
Dae-Jin
2007/1/13, Uwe Ligges [EMAIL PROTECTED] :
Dae-Jin Lee wrote:
Dear All
1)
I'm fitting spatial CAR models
using R2Winbugs and although everything seems to go reasonably well (or
I
think so)
the next message appears from WINBUGS 1.4 window:
gen.inits()
Command #Bugs: gen.inits cannot be executed (is greyed out)
greyed out refers to the corresponding GUI. In this case it could mean
that either your preparations before were not successful or you have
already specified all initial values and no further have to be
generated. But this is a WinBUGS question and not R related.
Perhaps you have transposed some matrix? This is always extremely
confusing ...
The question is if this message means that something is wrong and the
results are consequently wrong, or Can I assume it as a simple warning
message???...
I've tried to fit the model using just WinBugs (not calling form R) and
gen.inits command is available and the results obtain are practically
the
same...
2)
The other question is, once several bayesian CAR models are fitted I use
DIC
to model selection
How can I compare DIC with AIC or BIC obtain for a spatial glm? I mean
for
example:
- DIC of 300 and Pd of 32
versus
- BIC of 220 and Effective dimension (measured of trace of hat matrix)
equal
to 14
* I've read that DIC is intended as a generalisation of Akaike's
Information
Criterion (AIC), but is it possible to compare them simply looking which
is
the lowest???
Well they are intended to be compared (DIC with DIC, AIC with AIC, but
do not mix them!) that way given you are comparing within the same class
of models. And then, there is a bit of fortunetelling with all these
information criterion, at least in my personal opinion (and I know that
others disagree).
Uwe Ligges
Thanks,
Dae-Jin Lee
[[alternative HTML version deleted]]
__
R-help@stat.math.ethz.ch mailing list
https://stat.ethz.ch/mailman/listinfo/r-help
PLEASE do read the posting guide
http://www.R-project.org/posting-guide.html
and provide commented, minimal, self-contained, reproducible code.
[[alternative HTML version deleted]]
__
R-help@stat.math.ethz.ch mailing list
https://stat.ethz.ch/mailman/listinfo/r-help
PLEASE do read the posting guide http://www.R-project.org/posting-guide.html
and provide commented, minimal, self-contained, reproducible code.
Re: [R] R2WinBugs and Compare DIC versus BIC or AIC
Dae-Jin Lee wrote:
Thanks for your answer !
But If I have understood correctly, the procedure is correct and this only
means that I have already generated the initial values
From R. I define a function to generate the initial values as:
inits. - function(){list(sdesp=runif(1),sdhet=runif(1),beta0=runif(1),
beta1=runif(1), b=rnorm(n), phi=rnorm(n))}
then It is not necessary to generate initial values, that why
gen.inits() is
greyed out
, is it ok?
Yes, alright.
Uwe
Here is what Winbugs show
==
display(log)
check(D:/model.txt)
model is syntactically correct
data(D://data.txt)
data loaded
compile(3)
model compiled
inits(1,D:/inits1.txt)
chain initialized but other chain(s) contain uninitialized variables
inits(2,D:/inits2.txt)
chain initialized but other chain(s) contain uninitialized variables
inits(3,D:/inits3.txt)
model is initialized
gen.inits()
command #Bugs:gen.inits cannot be executed (is greyed out)
thin.updater(1)
update(5000)
==
Thank again in advance
Dae-Jin
2007/1/13, Uwe Ligges [EMAIL PROTECTED] :
Dae-Jin Lee wrote:
Dear All
1)
I'm fitting spatial CAR models
using R2Winbugs and although everything seems to go reasonably well (or
I
think so)
the next message appears from WINBUGS 1.4 window:
gen.inits()
Command #Bugs: gen.inits cannot be executed (is greyed out)
greyed out refers to the corresponding GUI. In this case it could mean
that either your preparations before were not successful or you have
already specified all initial values and no further have to be
generated. But this is a WinBUGS question and not R related.
Perhaps you have transposed some matrix? This is always extremely
confusing ...
The question is if this message means that something is wrong and the
results are consequently wrong, or Can I assume it as a simple warning
message???...
I've tried to fit the model using just WinBugs (not calling form R) and
gen.inits command is available and the results obtain are practically
the
same...
2)
The other question is, once several bayesian CAR models are fitted I
use
DIC
to model selection
How can I compare DIC with AIC or BIC obtain for a spatial glm? I mean
for
example:
- DIC of 300 and Pd of 32
versus
- BIC of 220 and Effective dimension (measured of trace of hat matrix)
equal
to 14
* I've read that DIC is intended as a generalisation of Akaike's
Information
Criterion (AIC), but is it possible to compare them simply looking
which
is
the lowest???
Well they are intended to be compared (DIC with DIC, AIC with AIC, but
do not mix them!) that way given you are comparing within the same class
of models. And then, there is a bit of fortunetelling with all these
information criterion, at least in my personal opinion (and I know that
others disagree).
Uwe Ligges
Thanks,
Dae-Jin Lee
[[alternative HTML version deleted]]
__
R-help@stat.math.ethz.ch mailing list
https://stat.ethz.ch/mailman/listinfo/r-help
PLEASE do read the posting guide
http://www.R-project.org/posting-guide.html
and provide commented, minimal, self-contained, reproducible code.
__
R-help@stat.math.ethz.ch mailing list
https://stat.ethz.ch/mailman/listinfo/r-help
PLEASE do read the posting guide http://www.R-project.org/posting-guide.html
and provide commented, minimal, self-contained, reproducible code.
Re: [R] R2WinBugs and Compare DIC versus BIC or AIC
Dae-Jin Lee wrote: Dear All 1) I'm fitting spatial CAR models using R2Winbugs and although everything seems to go reasonably well (or I think so) the next message appears from WINBUGS 1.4 window: gen.inits() Command #Bugs: gen.inits cannot be executed (is greyed out) greyed out refers to the corresponding GUI. In this case it could mean that either your preparations before were not successful or you have already specified all initial values and no further have to be generated. But this is a WinBUGS question and not R related. Perhaps you have transposed some matrix? This is always extremely confusing ... The question is if this message means that something is wrong and the results are consequently wrong, or Can I assume it as a simple warning message???... I've tried to fit the model using just WinBugs (not calling form R) and gen.inits command is available and the results obtain are practically the same... 2) The other question is, once several bayesian CAR models are fitted I use DIC to model selection How can I compare DIC with AIC or BIC obtain for a spatial glm? I mean for example: - DIC of 300 and Pd of 32 versus - BIC of 220 and Effective dimension (measured of trace of hat matrix) equal to 14 * I've read that DIC is intended as a generalisation of Akaike's Information Criterion (AIC), but is it possible to compare them simply looking which is the lowest??? Well they are intended to be compared (DIC with DIC, AIC with AIC, but do not mix them!) that way given you are comparing within the same class of models. And then, there is a bit of fortunetelling with all these information criterion, at least in my personal opinion (and I know that others disagree). Uwe Ligges Thanks, Dae-Jin Lee [[alternative HTML version deleted]] __ R-help@stat.math.ethz.ch mailing list https://stat.ethz.ch/mailman/listinfo/r-help PLEASE do read the posting guide http://www.R-project.org/posting-guide.html and provide commented, minimal, self-contained, reproducible code. __ R-help@stat.math.ethz.ch mailing list https://stat.ethz.ch/mailman/listinfo/r-help PLEASE do read the posting guide http://www.R-project.org/posting-guide.html and provide commented, minimal, self-contained, reproducible code.
[R] R2WinBugs and Compare DIC versus BIC or AIC
Dear All 1) I'm fitting spatial CAR models using R2Winbugs and although everything seems to go reasonably well (or I think so) the next message appears from WINBUGS 1.4 window: gen.inits() Command #Bugs: gen.inits cannot be executed (is greyed out) The question is if this message means that something is wrong and the results are consequently wrong, or Can I assume it as a simple warning message???... I've tried to fit the model using just WinBugs (not calling form R) and gen.inits command is available and the results obtain are practically the same... 2) The other question is, once several bayesian CAR models are fitted I use DIC to model selection How can I compare DIC with AIC or BIC obtain for a spatial glm? I mean for example: - DIC of 300 and Pd of 32 versus - BIC of 220 and Effective dimension (measured of trace of hat matrix) equal to 14 * I've read that DIC is intended as a generalisation of Akaike's Information Criterion (AIC), but is it possible to compare them simply looking which is the lowest??? Thanks, Dae-Jin Lee [[alternative HTML version deleted]] __ R-help@stat.math.ethz.ch mailing list https://stat.ethz.ch/mailman/listinfo/r-help PLEASE do read the posting guide http://www.R-project.org/posting-guide.html and provide commented, minimal, self-contained, reproducible code.
Re: [R] R2WinBUGS and calling WinBUGS in Crossover MAC BETA
Ben Bolker wrote:
Kenneth Benoit kbenoit at tcd.ie writes:
Hi there - I have a question for any of you who use R2WinBUGS to call
WinBUGS using the useWINE option in that package.
I have Codeweaver's Crossover emulator installed on my Intel Mac with
WinBUGS working fine if I start it from Crossover. But there is
supposed to be a way to start it directly from a command line, which I
could pass to the R2WinBUGS using the bugs(..., useWINE=something)
argument.
Has anyone tried this yet? Thanks!
useWINE=TRUE. You may have to set some environment
variables (e.g. WINE is the path to your WINE executable).
It's supposed to just work out of the box. I have some
fixes that I've been meaning to upload to complement
the existing stuff, but haven't. If you run into
trouble drop me a line.
Ben Bolker
Many thanks for that quick reply. I have tried useWINE=TRUE but this
does not work on the Mac. I am using CrossOver for Mac Beta (6.0.0b2).
When I call:
bugs(..., useWINE=TRUE, newWINE=TRUE)
then I get this message:
Error in if (!nchar(WINEPATH)) { : argument is of length zero
I think the problem is that the command line call for cxoffice (similar
to wine) is different on the Mac from the Linux version. For instance
I have the following application bundles:
/Applications/CrossOver.app/
~/Applications/CrossOver/WinBUGS14.app
but I cannot figure out how to start WinBUGS14.app from the command
line. If I did then I could supply this command to bugs(..., WINE=).
Best,
Ken
__
R-help@stat.math.ethz.ch mailing list
https://stat.ethz.ch/mailman/listinfo/r-help
PLEASE do read the posting guide http://www.R-project.org/posting-guide.html
and provide commented, minimal, self-contained, reproducible code.
[R] R2WinBUGS - parameter monitor seting
R2WinBUGS users,
I'm getting the error message monitor could not be set when I try to
monitor (and output to R) a response variable that includes missing data
(NAs)...ie.e imputing for y. Setting monitors for other parameters works
OK and running the R-generated WinBUGS script file in WinBUGS, also
works OK. Below is my R code and the log.txt file output by R2WinBUGS.
As I want to run this model 100+ times, i'd prefer to write some looping
code rather than cut and paste 100 times..can anyone help?
thanks
doug
R-code:
bayesimp.sim - bugs(data, inits, parameters,bayesimp.bug,
n.chains = 1, n.iter = 4000,
bugs.directory = c:/Program Files/WinBUGS14/,
working.directory = NULL, clearWD = TRUE, debug = TRUE)
script.txt:
display ('log')
check ('D:/Work/SpatialEpi/Water/THM Imputation/Rwork/bayesimp.txt')
data ('D:/Work/SpatialEpi/Water/THM Imputation/Rwork/data.txt')
compile (1)
inits (1, 'D:/Work/SpatialEpi/Water/THM Imputation/Rwork/inits1.txt')
gen.inits()
thin.updater (2)
update (1000)
set (y)
set (u)
set (deviance)
dic.set()
update (1000)
coda (*, 'D:/Work/SpatialEpi/Water/THM Imputation/Rwork/coda')
stats (*)
dic.stats()
history (*, 'D:/Work/SpatialEpi/Water/THM
Imputation/Rwork/history.odc')
save ('D:/Work/SpatialEpi/Water/THM Imputation/Rwork/log.odc')
save ('D:/Work/SpatialEpi/Water/THM Imputation/Rwork/log.txt')
log.txt:
display(log)
check(D:/Work/SpatialEpi/Water/THM Imputation/Rwork/bayesimp.txt)
model is syntactically correct
data(D:/Work/SpatialEpi/Water/THM Imputation/Rwork/data.txt)
data loaded
compile(1)
model compiled
inits(1,D:/Work/SpatialEpi/Water/THM Imputation/Rwork/inits1.txt)
this chain contains uninitialized variables
gen.inits()
initial values generated, model initialized
thin.updater(2)
update(1000)
set(y)
monitor could not be set
set(u)
set(deviance)
dic.set()
update(1000)
coda(*,D:/Work/SpatialEpi/Water/THM Imputation/Rwork/coda)
stats(*)
This email and any files transmitted with it are confidentia...{{dropped}}
__
R-help@stat.math.ethz.ch mailing list
https://stat.ethz.ch/mailman/listinfo/r-help
PLEASE do read the posting guide http://www.R-project.org/posting-guide.html
and provide commented, minimal, self-contained, reproducible code.
Re: [R] R2WinBugs
We had set R2WinBugs to use the burnin as the adaptive phase. I think this was changed very slightly in the latest version of R2WinBUGS so that the adaptive phase would equal burnin minus 1. This was to allow DIC to be calculated. So I think it should work OK now. If you send an example I can take a look. Andrew Uwe Ligges wrote: Bowden, J.M. wrote: Hi all, I am having problems using the R2Winbugs function When I perform an analysis directly in Winbugs I can specify that the first 'n' iterations are to be done using an 'adaptive' phase. After this phase the markov chain seems to mix a lot better. I don't seem to be able to specify R2winbugs to carry out this adaptive phase, I can just specify the burnin length but this (to my knowledge) is not the same thing. As a result my models are not fitting as well as I would like, has anyone had a similar experience? I think this is a question for Andrew Gelman (CCing), who wrote the underlying code for automatically setting the adaptive phase. Uwe Ligges Jack __ R-help@stat.math.ethz.ch mailing list https://stat.ethz.ch/mailman/listinfo/r-help PLEASE do read the posting guide http://www.R-project.org/posting-guide.html and provide commented, minimal, self-contained, reproducible code. -- Andrew Gelman Professor, Department of Statistics Professor, Department of Political Science [EMAIL PROTECTED] www.stat.columbia.edu/~gelman Statistics department office: Social Work Bldg (Amsterdam Ave at 122 St), Room 1016 212-851-2142 Political Science department office: International Affairs Bldg (Amsterdam Ave at 118 St), Room 731 212-854-7075 Mailing address: 1255 Amsterdam Ave, Room 1016 Columbia University New York, NY 10027-5904 212-851-2142 (fax) 212-851-2164 __ R-help@stat.math.ethz.ch mailing list https://stat.ethz.ch/mailman/listinfo/r-help PLEASE do read the posting guide http://www.R-project.org/posting-guide.html and provide commented, minimal, self-contained, reproducible code.
[R] R2WinBugs
Hi all, I am having problems using the R2Winbugs function When I perform an analysis directly in Winbugs I can specify that the first 'n' iterations are to be done using an 'adaptive' phase. After this phase the markov chain seems to mix a lot better. I don't seem to be able to specify R2winbugs to carry out this adaptive phase, I can just specify the burnin length but this (to my knowledge) is not the same thing. As a result my models are not fitting as well as I would like, has anyone had a similar experience? Jack __ R-help@stat.math.ethz.ch mailing list https://stat.ethz.ch/mailman/listinfo/r-help PLEASE do read the posting guide http://www.R-project.org/posting-guide.html and provide commented, minimal, self-contained, reproducible code.
Re: [R] R2WinBugs
Bowden, J.M. wrote: Hi all, I am having problems using the R2Winbugs function When I perform an analysis directly in Winbugs I can specify that the first 'n' iterations are to be done using an 'adaptive' phase. After this phase the markov chain seems to mix a lot better. I don't seem to be able to specify R2winbugs to carry out this adaptive phase, I can just specify the burnin length but this (to my knowledge) is not the same thing. As a result my models are not fitting as well as I would like, has anyone had a similar experience? I think this is a question for Andrew Gelman (CCing), who wrote the underlying code for automatically setting the adaptive phase. Uwe Ligges Jack __ R-help@stat.math.ethz.ch mailing list https://stat.ethz.ch/mailman/listinfo/r-help PLEASE do read the posting guide http://www.R-project.org/posting-guide.html and provide commented, minimal, self-contained, reproducible code. __ R-help@stat.math.ethz.ch mailing list https://stat.ethz.ch/mailman/listinfo/r-help PLEASE do read the posting guide http://www.R-project.org/posting-guide.html and provide commented, minimal, self-contained, reproducible code.
Re: [R] R2WinBUGS error
Joseph Retzer wrote:
Dear R-help,
I'm using the R2WinBUGS package and getting an error message:
Error in file(file, r) : unable to open connection
In addition: Warning message:
cannot open file 'codaIndex.txt', reason 'No such file or
directory'
Looks like there was an error in WinBUGS: R cannot access the file,
because WinBUGS has not written it before.
You can use
bugs(.., debug = TRUE)
Then WinBUGS stays open and you can look at its error messages in order
to correct the code/data/inits or whatever
Uwe Ligges
I'm using R 2.2.1 and WinBUGS 1.4.1 on a windows machine (XP). My R code
and WinBUGS code is given below. The complete WinBUGS program executes
correctly in WinBUGS however I'm generating some of my inits using WinBUGS
so I may be making an error there. On the other hand, the error generated in
R seems to imply it cannot locate a file. I've checked my paths and they are
correct. Also, my data is loading correctly.
Many thanks,
Joe
#
# R code
# Runs Bayesian Ordered Logit by calling WinBUGS from R
# ologit2.txt: WinBUGS commands
library(R2WinBUGS)
setwd(c:/docume~1/admini~1/mydocu~1/r_tuto~1)
load(oldat.Rdata) # R data file containing data frame ol.dat
# with vars: q02, bf23f, bf23b, bf22, bf34a, bf34.1,
bf34.2
q02 -ol.dat$q02
bf23f - ol.dat$bf23f
bf23b - ol.dat$bf23b
bf22 - ol.dat$bf22
bf34a - ol.dat$bf34a
bf34.1 - ol.dat$bf34.1
bf34.2 - ol.dat$bf34.2
N=nrow(ol.dat)
Ncut=5
data - list(N, q02, bf23f, bf23b, bf22, bf34a, bf34.1,
bf34.2, Ncut)
inits - function()
{
list(k=c(-5, -4, -3, -2, -1), tau=2, theta=rnorm(7, -1, 100))
}
parameters - c(k)
olog.out - bugs(data, inits, parameters,
model.file=c:/Documents and Settings/Administrator/My
Documents/r_tutorial/ologit2.txt,
n.chains = 2, n.iter = 1000,
bugs.directory = c:/Program Files/WinBUGS14/)
# WinBUGS code
model exec;
{
# Priors on regression coefficients
theta[1] ~ dnorm( -1,1.0) ; theta[2] ~ dnorm(-1,1.0)
theta[3] ~ dnorm( 1,1.0) ; theta[4] ~ dnorm(-1,1.0)
theta[5] ~ dnorm( -1,1.0) ; theta[6] ~ dnorm( 1,1.0)
theta[7] ~ dnorm( -1,1.0)
# Priors on latent variable cutpoints
k[1] ~ dnorm(0, 0.1)I(, k[2]); k[2] ~ dnorm(0, 0.1)I(k[1], k[3])
k[3] ~ dnorm(0, 0.1)I(k[2], k[4]); k[4] ~ dnorm(0, 0.1)I(k[3], k[5])
k[5] ~ dnorm(0, 0.1)I(k[4], )
# Prior on precision
tau ~ dgamma(0.001, 0.001)
# Some defs
sigma - sqrt(1 / tau); log.sigma - log(sigma);
for (i in 1 : N)
{
# Prior on
b[i] ~ dnorm(0.0, tau)
# Model Mean
mu[i] - theta[1] + theta[2]*bf22[i] + theta[3]*bf23b[i] +
theta[4]*bf23f[i] + theta[5]*bf34a[i] + theta[6]*bf34.1[i] +
theta[7]*bf34.2[i]
for (j in 1 : Ncut)
{
# Logit Model
# cumulative probability of lower response than j
logit(Q[i, j]) - -(k[j] + mu[i] + b[i])
}
# probability that response = j
p[i,1] - max( min(1 - Q[i,1], 1), 0)
for (j in 2 : Ncut)
{
p[i,j] - max( min(Q[i,j-1] - Q[i,j],1), 0)
}
p[i,(Ncut+1)] - max( min(Q[i,Ncut], 1), 0)
q02[i] ~ dcat(p[i, ])
}
}
[[alternative HTML version deleted]]
__
R-help@stat.math.ethz.ch mailing list
https://stat.ethz.ch/mailman/listinfo/r-help
PLEASE do read the posting guide! http://www.R-project.org/posting-guide.html
__
R-help@stat.math.ethz.ch mailing list
https://stat.ethz.ch/mailman/listinfo/r-help
PLEASE do read the posting guide! http://www.R-project.org/posting-guide.html
Re: [R] R2WinBUGS error Solved, THANK YOU!
Thanks very much Uwe! There was a problem reading my data which I would not
have discovered without careful examination of the WinBUGS log file. The
program is working now!
Take care,
Joe
-Original Message-
From: Uwe Ligges [mailto:[EMAIL PROTECTED]
Sent: Thursday, April 06, 2006 1:30 AM
To: Joseph Retzer
Cc: r-help@stat.math.ethz.ch
Subject: Re: [R] R2WinBUGS error
Joseph Retzer wrote:
Dear R-help,
I'm using the R2WinBUGS package and getting an error message:
Error in file(file, r) : unable to open connection
In addition: Warning message:
cannot open file 'codaIndex.txt', reason 'No such file or
directory'
Looks like there was an error in WinBUGS: R cannot access the file, because
WinBUGS has not written it before.
You can use
bugs(.., debug = TRUE)
Then WinBUGS stays open and you can look at its error messages in order to
correct the code/data/inits or whatever
Uwe Ligges
I'm using R 2.2.1 and WinBUGS 1.4.1 on a windows machine (XP). My R
code and WinBUGS code is given below. The complete WinBUGS program
executes correctly in WinBUGS however I'm generating some of my inits
using WinBUGS so I may be making an error there. On the other hand,
the error generated in R seems to imply it cannot locate a file. I've
checked my paths and they are correct. Also, my data is loading correctly.
Many thanks,
Joe
#
# R code
# Runs Bayesian Ordered Logit by calling WinBUGS from R
# ologit2.txt: WinBUGS commands
library(R2WinBUGS)
setwd(c:/docume~1/admini~1/mydocu~1/r_tuto~1)
load(oldat.Rdata) # R data file containing data frame ol.dat
# with vars: q02, bf23f, bf23b, bf22, bf34a,
bf34.1,
bf34.2
q02 -ol.dat$q02
bf23f - ol.dat$bf23f
bf23b - ol.dat$bf23b
bf22 - ol.dat$bf22
bf34a - ol.dat$bf34a
bf34.1 - ol.dat$bf34.1
bf34.2 - ol.dat$bf34.2
N=nrow(ol.dat)
Ncut=5
data - list(N, q02, bf23f, bf23b, bf22, bf34a, bf34.1,
bf34.2, Ncut)
inits - function()
{
list(k=c(-5, -4, -3, -2, -1), tau=2, theta=rnorm(7, -1, 100))
}
parameters - c(k)
olog.out - bugs(data, inits, parameters,
model.file=c:/Documents and
Settings/Administrator/My Documents/r_tutorial/ologit2.txt,
n.chains = 2, n.iter = 1000,
bugs.directory = c:/Program Files/WinBUGS14/)
# WinBUGS code
model exec;
{
# Priors on regression coefficients
theta[1] ~ dnorm( -1,1.0) ; theta[2] ~ dnorm(-1,1.0)
theta[3] ~ dnorm( 1,1.0) ; theta[4] ~ dnorm(-1,1.0)
theta[5] ~ dnorm( -1,1.0) ; theta[6] ~ dnorm( 1,1.0)
theta[7] ~ dnorm( -1,1.0)
# Priors on latent variable cutpoints
k[1] ~ dnorm(0, 0.1)I(, k[2]); k[2] ~ dnorm(0, 0.1)I(k[1], k[3])
k[3] ~ dnorm(0, 0.1)I(k[2], k[4]); k[4] ~ dnorm(0, 0.1)I(k[3], k[5])
k[5] ~ dnorm(0, 0.1)I(k[4], )
# Prior on precision
tau ~ dgamma(0.001, 0.001)
# Some defs
sigma - sqrt(1 / tau); log.sigma - log(sigma);
for (i in 1 : N)
{
# Prior on
b[i] ~ dnorm(0.0, tau)
# Model Mean
mu[i] - theta[1] + theta[2]*bf22[i] + theta[3]*bf23b[i] +
theta[4]*bf23f[i] + theta[5]*bf34a[i] + theta[6]*bf34.1[i] +
theta[7]*bf34.2[i]
for (j in 1 : Ncut)
{
# Logit Model
# cumulative probability of lower response than j
logit(Q[i, j]) - -(k[j] + mu[i] + b[i])
}
# probability that response = j
p[i,1] - max( min(1 - Q[i,1], 1), 0)
for (j in 2 : Ncut)
{
p[i,j] - max( min(Q[i,j-1] - Q[i,j],1), 0)
}
p[i,(Ncut+1)] - max( min(Q[i,Ncut], 1), 0)
q02[i] ~ dcat(p[i, ])
}
}
[[alternative HTML version deleted]]
__
R-help@stat.math.ethz.ch mailing list
https://stat.ethz.ch/mailman/listinfo/r-help
PLEASE do read the posting guide!
http://www.R-project.org/posting-guide.html
__
R-help@stat.math.ethz.ch mailing list
https://stat.ethz.ch/mailman/listinfo/r-help
PLEASE do read the posting guide! http://www.R-project.org/posting-guide.html
[R] R2WinBUGS error
Dear R-help,
I'm using the R2WinBUGS package and getting an error message:
Error in file(file, r) : unable to open connection
In addition: Warning message:
cannot open file 'codaIndex.txt', reason 'No such file or
directory'
I'm using R 2.2.1 and WinBUGS 1.4.1 on a windows machine (XP). My R code
and WinBUGS code is given below. The complete WinBUGS program executes
correctly in WinBUGS however I'm generating some of my inits using WinBUGS
so I may be making an error there. On the other hand, the error generated in
R seems to imply it cannot locate a file. I've checked my paths and they are
correct. Also, my data is loading correctly.
Many thanks,
Joe
#
# R code
# Runs Bayesian Ordered Logit by calling WinBUGS from R
# ologit2.txt: WinBUGS commands
library(R2WinBUGS)
setwd(c:/docume~1/admini~1/mydocu~1/r_tuto~1)
load(oldat.Rdata) # R data file containing data frame ol.dat
# with vars: q02, bf23f, bf23b, bf22, bf34a, bf34.1,
bf34.2
q02 -ol.dat$q02
bf23f - ol.dat$bf23f
bf23b - ol.dat$bf23b
bf22 - ol.dat$bf22
bf34a - ol.dat$bf34a
bf34.1 - ol.dat$bf34.1
bf34.2 - ol.dat$bf34.2
N=nrow(ol.dat)
Ncut=5
data - list(N, q02, bf23f, bf23b, bf22, bf34a, bf34.1,
bf34.2, Ncut)
inits - function()
{
list(k=c(-5, -4, -3, -2, -1), tau=2, theta=rnorm(7, -1, 100))
}
parameters - c(k)
olog.out - bugs(data, inits, parameters,
model.file=c:/Documents and Settings/Administrator/My
Documents/r_tutorial/ologit2.txt,
n.chains = 2, n.iter = 1000,
bugs.directory = c:/Program Files/WinBUGS14/)
# WinBUGS code
model exec;
{
# Priors on regression coefficients
theta[1] ~ dnorm( -1,1.0) ; theta[2] ~ dnorm(-1,1.0)
theta[3] ~ dnorm( 1,1.0) ; theta[4] ~ dnorm(-1,1.0)
theta[5] ~ dnorm( -1,1.0) ; theta[6] ~ dnorm( 1,1.0)
theta[7] ~ dnorm( -1,1.0)
# Priors on latent variable cutpoints
k[1] ~ dnorm(0, 0.1)I(, k[2]); k[2] ~ dnorm(0, 0.1)I(k[1], k[3])
k[3] ~ dnorm(0, 0.1)I(k[2], k[4]); k[4] ~ dnorm(0, 0.1)I(k[3], k[5])
k[5] ~ dnorm(0, 0.1)I(k[4], )
# Prior on precision
tau ~ dgamma(0.001, 0.001)
# Some defs
sigma - sqrt(1 / tau); log.sigma - log(sigma);
for (i in 1 : N)
{
# Prior on
b[i] ~ dnorm(0.0, tau)
# Model Mean
mu[i] - theta[1] + theta[2]*bf22[i] + theta[3]*bf23b[i] +
theta[4]*bf23f[i] + theta[5]*bf34a[i] + theta[6]*bf34.1[i] +
theta[7]*bf34.2[i]
for (j in 1 : Ncut)
{
# Logit Model
# cumulative probability of lower response than j
logit(Q[i, j]) - -(k[j] + mu[i] + b[i])
}
# probability that response = j
p[i,1] - max( min(1 - Q[i,1], 1), 0)
for (j in 2 : Ncut)
{
p[i,j] - max( min(Q[i,j-1] - Q[i,j],1), 0)
}
p[i,(Ncut+1)] - max( min(Q[i,Ncut], 1), 0)
q02[i] ~ dcat(p[i, ])
}
}
[[alternative HTML version deleted]]
__
R-help@stat.math.ethz.ch mailing list
https://stat.ethz.ch/mailman/listinfo/r-help
PLEASE do read the posting guide! http://www.R-project.org/posting-guide.html
Re: [R] R2WinBUGS error
Dear R-help,
I'm using the R2WinBUGS package and getting an error message:
Error in file(file, r) : unable to open connection
In addition: Warning message:
cannot open file 'codaIndex.txt', reason 'No such file or
directory'
I'm using R 2.2.1 and WinBUGS 1.4.1 on a windows machine (XP). My R code
and WinBUGS code is given below. The complete WinBUGS program executes
correctly in WinBUGS however I'm generating some of my inits using WinBUGS
so I may be making an error there. On the other hand, the error generated in
R seems to imply it cannot locate a file. I've checked my paths and they are
correct. Also, my data is loading correctly.
Many thanks,
Joe
#
# R code
# Runs Bayesian Ordered Logit by calling WinBUGS from R
# ologit2.txt: WinBUGS commands
library(R2WinBUGS)
setwd(c:/docume~1/admini~1/mydocu~1/r_tuto~1)
load(oldat.Rdata) # R data file containing data frame ol.dat
# with vars: q02, bf23f, bf23b, bf22, bf34a, bf34.1,
bf34.2
q02 -ol.dat$q02
bf23f - ol.dat$bf23f
bf23b - ol.dat$bf23b
bf22 - ol.dat$bf22
bf34a - ol.dat$bf34a
bf34.1 - ol.dat$bf34.1
bf34.2 - ol.dat$bf34.2
N=nrow(ol.dat)
Ncut=5
data - list(N, q02, bf23f, bf23b, bf22, bf34a, bf34.1,
bf34.2, Ncut)
inits - function()
{
list(k=c(-5, -4, -3, -2, -1), tau=2, theta=rnorm(7, -1, 100))
}
parameters - c(k)
olog.out - bugs(data, inits, parameters,
model.file=c:/Documents and Settings/Administrator/My
Documents/r_tutorial/ologit2.txt,
n.chains = 2, n.iter = 1000,
bugs.directory = c:/Program Files/WinBUGS14/)
# WinBUGS code
model exec;
{
# Priors on regression coefficients
theta[1] ~ dnorm( -1,1.0) ; theta[2] ~ dnorm(-1,1.0)
theta[3] ~ dnorm( 1,1.0) ; theta[4] ~ dnorm(-1,1.0)
theta[5] ~ dnorm( -1,1.0) ; theta[6] ~ dnorm( 1,1.0)
theta[7] ~ dnorm( -1,1.0)
# Priors on latent variable cutpoints
k[1] ~ dnorm(0, 0.1)I(, k[2]); k[2] ~ dnorm(0, 0.1)I(k[1], k[3])
k[3] ~ dnorm(0, 0.1)I(k[2], k[4]); k[4] ~ dnorm(0, 0.1)I(k[3], k[5])
k[5] ~ dnorm(0, 0.1)I(k[4], )
# Prior on precision
tau ~ dgamma(0.001, 0.001)
# Some defs
sigma - sqrt(1 / tau); log.sigma - log(sigma);
for (i in 1 : N)
{
# Prior on
b[i] ~ dnorm(0.0, tau)
# Model Mean
mu[i] - theta[1] + theta[2]*bf22[i] + theta[3]*bf23b[i] +
theta[4]*bf23f[i] + theta[5]*bf34a[i] + theta[6]*bf34.1[i] +
theta[7]*bf34.2[i]
for (j in 1 : Ncut)
{
# Logit Model
# cumulative probability of lower response than j
logit(Q[i, j]) - -(k[j] + mu[i] + b[i])
}
# probability that response = j
p[i,1] - max( min(1 - Q[i,1], 1), 0)
for (j in 2 : Ncut)
{
p[i,j] - max( min(Q[i,j-1] - Q[i,j],1), 0)
}
p[i,(Ncut+1)] - max( min(Q[i,Ncut], 1), 0)
q02[i] ~ dcat(p[i, ])
}
}
[[alternative HTML version deleted]]
__
R-help@stat.math.ethz.ch mailing list
https://stat.ethz.ch/mailman/listinfo/r-help
PLEASE do read the posting guide! http://www.R-project.org/posting-guide.html
[R] R2WinBUGS erro
Dear R-help,
I'm using the R2WinBUGS package and getting an error message:
Error in file(file, r) : unable to open connection
In addition: Warning message:
cannot open file 'codaIndex.txt', reason 'No such file or directory'
I'm using R 2.2.1 and WinBUGS 1.4.1 on a windows machine (XP). My R code and
WinBUGS code is given below. The complete WinBUGS program executes correctly in
WinBUGS however I'm generating some of my inits using WinBUGS so I may be
making an error there. On the other hand, the error generated in R seems to
imply it cannot locate a file. I've checked my paths and they are correct.
Also, my data is loading correctly.
Many thanks,
Joe
#
# R code
# Runs Bayesian Ordered Logit by calling WinBUGS from R
# ologit2.txt: WinBUGS commands
library(R2WinBUGS)
setwd(c:/docume~1/admini~1/mydocu~1/r_tuto~1)
load(oldat.Rdata) # R data file containing data frame ol.dat
# with vars: q02, bf23f, bf23b, bf22, bf34a, bf34.1, bf34.2
q02 -ol.dat$q02
bf23f - ol.dat$bf23f
bf23b - ol.dat$bf23b
bf22 - ol.dat$bf22
bf34a - ol.dat$bf34a
bf34.1 - ol.dat$bf34.1
bf34.2 - ol.dat$bf34.2
N=nrow(ol.dat)
Ncut=5
data - list(N, q02, bf23f, bf23b, bf22, bf34a, bf34.1,
bf34.2, Ncut)
inits - function()
{
list(k=c(-5, -4, -3, -2, -1), tau=2, theta=rnorm(7, -1, 100))
}
parameters - c(k)
olog.out - bugs(data, inits, parameters,
model.file=c:/Documents and Settings/Administrator/My
Documents/r_tutorial/ologit2.txt,
n.chains = 2, n.iter = 1000,
bugs.directory = c:/Program Files/WinBUGS14/)
# WinBUGS code
model exec;
{
# Priors on regression coefficients
theta[1] ~ dnorm( -1,1.0) ; theta[2] ~ dnorm(-1,1.0)
theta[3] ~ dnorm( 1,1.0) ; theta[4] ~ dnorm(-1,1.0)
theta[5] ~ dnorm( -1,1.0) ; theta[6] ~ dnorm( 1,1.0)
theta[7] ~ dnorm( -1,1.0)
# Priors on latent variable cutpoints
k[1] ~ dnorm(0, 0.1)I(, k[2]); k[2] ~ dnorm(0, 0.1)I(k[1], k[3])
k[3] ~ dnorm(0, 0.1)I(k[2], k[4]); k[4] ~ dnorm(0, 0.1)I(k[3], k[5])
k[5] ~ dnorm(0, 0.1)I(k[4], )
# Prior on precision
tau ~ dgamma(0.001, 0.001)
# Some defs
sigma - sqrt(1 / tau); log.sigma - log(sigma);
for (i in 1 : N)
{
# Prior on
b[i] ~ dnorm(0.0, tau)
# Model Mean
mu[i] - theta[1] + theta[2]*bf22[i] + theta[3]*bf23b[i] +
theta[4]*bf23f[i] + theta[5]*bf34a[i] + theta[6]*bf34.1[i] + theta[7]*bf34.2[i]
for (j in 1 : Ncut)
{
# Logit Model
# cumulative probability of lower response than j
logit(Q[i, j]) - -(k[j] + mu[i] + b[i])
}
# probability that response = j
p[i,1] - max( min(1 - Q[i,1], 1), 0)
for (j in 2 : Ncut)
{
p[i,j] - max( min(Q[i,j-1] - Q[i,j],1), 0)
}
p[i,(Ncut+1)] - max( min(Q[i,Ncut], 1), 0)
q02[i] ~ dcat(p[i, ])
}
}
[[alternative HTML version deleted]]
__
R-help@stat.math.ethz.ch mailing list
https://stat.ethz.ch/mailman/listinfo/r-help
PLEASE do read the posting guide! http://www.R-project.org/posting-guide.html
Re: [R] R2WinBUGS - formating data for winbugs
Chris Behr wrote: I am currently running analyses in winbugs with two different formats of data: matrix and list. The data in the list have different dimensions than the matrix. Do I need to create a single format for entry into R which R passes to WinBUGS? It is not quite clear to me what you mean, you have to be more precise and give some code example (possibly in a private mail to Sibylle Sturtz or me). I guess you need to have to types of objects inR that you want to pass to WinBUGS, in this case, simply specify a list of those objects as the data argument to the bugs() function. Uwe Ligges Thanks, Chris Christopher Behr Principal Analyst eDesign Dynamics www.edesigndynamics.com 4024 Calvert St. NW Washington DC 20007 (202) 298-6437 (t/f) (551) 998-4823 (c) __ R-help@stat.math.ethz.ch mailing list https://stat.ethz.ch/mailman/listinfo/r-help PLEASE do read the posting guide! http://www.R-project.org/posting-guide.html __ R-help@stat.math.ethz.ch mailing list https://stat.ethz.ch/mailman/listinfo/r-help PLEASE do read the posting guide! http://www.R-project.org/posting-guide.html
[R] R2WinBUGS - formating data for winbugs
I am currently running analyses in winbugs with two different formats of data: matrix and list. The data in the list have different dimensions than the matrix. Do I need to create a single format for entry into R which R passes to WinBUGS? Thanks, Chris Christopher Behr Principal Analyst eDesign Dynamics www.edesigndynamics.com 4024 Calvert St. NW Washington DC 20007 (202) 298-6437 (t/f) (551) 998-4823 (c) __ R-help@stat.math.ethz.ch mailing list https://stat.ethz.ch/mailman/listinfo/r-help PLEASE do read the posting guide! http://www.R-project.org/posting-guide.html
[R] R2WinBUGS: Comparison to WinBUGS
Hi R-Users! I know I posted the question before (see archives) but I have not been able to find the mistake. Again using R2WinBUGS and WinBUGS does not yield the same result (although to my opinion the commands are the same). The variance of the parameters is much bigger and the parameter estimates are a bit different, too. If anybody has the time and interest to get the files and data and see if he/she has the same problems, I would be happy to provide these (and to discuss it further). It seems though that it must be my mistake SOMEWHERE because nobody else reported any problems yet... Thanks. Hadassa -- Hadassa Brunschwig Birmannsgasse 10A CH-4055 Basel Switzerland Phone: +41 78 797 6065 Email: [EMAIL PROTECTED] __ R-help@stat.math.ethz.ch mailing list https://stat.ethz.ch/mailman/listinfo/r-help PLEASE do read the posting guide! http://www.R-project.org/posting-guide.html
Re: [R] R2WinBUGS: Comparison to WinBUGS
Hadassa Brunschwig wrote: Hi R-Users! I know I posted the question before (see archives) but I have not been able to find the mistake. Again using R2WinBUGS and WinBUGS does not yield the same result (although to my opinion the commands are the same). The variance of the parameters is much bigger and the parameter estimates are a bit different, too. If anybody has the time and interest to get the files and data and see if he/she has the same problems, I would be happy to provide these (and to discuss it further). It seems though that it must be my mistake SOMEWHERE because nobody else reported any problems yet... Thanks. Hadassa Please send me your data in a private message. Perhaps we (let me include Sibylle Sturtz here, since she certainly knows better) find some time during the next week to look into it. Uwe Ligges __ R-help@stat.math.ethz.ch mailing list https://stat.ethz.ch/mailman/listinfo/r-help PLEASE do read the posting guide! http://www.R-project.org/posting-guide.html
[R] R2WinBUGS: Comparison to WinBUGS
Hi R-Help!
I used R2WinBUGS and WinBUGS directly on the same model just to compare. It
seems I am still making a mistake: after running the function bugs() I tried to
plot the posteriors of the parameters by using read.bugs() to convert the output
to an mcmc object and then plot.mcmc() to plot the densities. Using the same
model, the same number of iterations, the same initial values and the same data
I get completely different plots for the densities (e.g. the range of one
parameter in R2WinBUGS is from 0 to 8 but in WinBUGS only from 1.5 to 3)??? That
means my results are different, too.
Also, on the plot it says N=345 which is not what I specified in the bugs()
function (I specified 12000 iterations).
Below I put some of the code I used (if that helps):
parameters -
c(tau,C0,st90,C0.pop,st90.pop,tau.cpop,tau.stpop,st90.pop80)
inits - inits - function(){
list(tau = rep(1, 17),tau.cpop = 0.2, tau.stpop = 1)
}
mcmcA -
bugs(dataA,inits,parameters,modelA,n.chains=3,debug=T,n.iter=12000,n.burnin=2001,
bugs.directory=c:/Program
Files/WinBUGS14,working.directory=C:/Documents and
Settings/Daikon/Roche/R2WinBUGS Output,codaPkg=T)
codaA1 - read.bugs(mcmcA[1])
plot(codaA1)
THANKS A LOT!!
--
Hadassa Brunschwig
Birmannsgasse 10A
CH-4055 Basel
Switzerland
Phone: +41 78 797 6065
Email: [EMAIL PROTECTED]
__
R-help@stat.math.ethz.ch mailing list
https://stat.ethz.ch/mailman/listinfo/r-help
PLEASE do read the posting guide! http://www.R-project.org/posting-guide.html
Re: [R] R2WinBUGS: Data loading error
Hadassa Brunschwig wrote:
Hi R-Help!
I am trying to use R2WinBUGS but I get the following error message in WinBUGS
(and there must be something wrong with my R statement as I tried it directly
in
WinBUGS and it worked):
display(log)
check(C:/Documents and Settings/Daikon/Roche/pop_model.txt)
model is syntactically correct
data(C:/Documents and Settings/Daikon/Roche/data.txt)
expected key word structure
compile(7)
...(and of course nothing works after that)
and when I close WinBUGS i get:
Error in file(file, r) : unable to open connection
In addition: Warning message:
cannot open file 'codaIndex.txt'
Does anyone know what this 'expected key word structure' means?
This is my R code (and I guess my model file is ok):
modelA - c(C:/Documents and Settings/Daikon/Roche/pop_model.txt)
n - length(unique(subsetA$subject))#number of subjects
nt - 13 #number of days
Y - subsetA$concentr #concentration per
day/subject
t - 1:13 #days
dataA - list(n,nt,Y,t)
parameters - c(tau,C0,st90,C0.pop,st90.pop,tau.cpop,tau.stpop)
inits -
list(tau=c(1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1),C0=5,st90=4,C0.pop=5,st90.pop=4,tau.cpop=0.2,tau.stpop=1)
mcmcA -
bugs(dataA,inits,parameters,modelA,debug=T,n.chains=7,bugs.directory=c:/Program
Files/WinBUGS14,working.directory=C:/Documents and Settings/Daikon/Roche)
Which versions of R, WinBUGS and R2WinBUGS?
Your example is not reproducible for us.
I'd take a look whether dimensions are OK and whether subsetA$concentr
is in appropriate object, but without data and model file I am unable to
help for the data part.
For the inits part, please see ?bugs:
inits: a list with n.chains elements; each element of the list is itself
a list of starting values for the WinBUGS model, or a function creating
(possibly random) initial values. Alternatively, if inits = NULL,
initial values are generated by WinBUGS
Looks like you want to have the same inits for each chain. In order not
to repeat the inits 7 times, you might want to specify them simply as a
function such as:
inits - function(){
list(tau = rep(1, 17), C0 = 5, st90 = 4, C0.pop = 5, st90.pop = 4,
tau.cpop = 0.2, tau.stpop = 1)
}
Uwe Ligges
[Further correspondence on this particular topic, please respond to to
the package maintainer (Sibylle) and me directly rather than to R-help.]
Thanks so much...
Hadassa
__
R-help@stat.math.ethz.ch mailing list
https://stat.ethz.ch/mailman/listinfo/r-help
PLEASE do read the posting guide! http://www.R-project.org/posting-guide.html
Re: [R] R2WinBUGS: Data loading error
Thank you for your answer. I found out the mistake and it finally works!!
Hadassa
--
Hadassa Brunschwig
Birmannsgasse 10A
CH-4055 Basel
Switzerland
Phone: +41 78 797 6065
Email: [EMAIL PROTECTED]
Quoting Uwe Ligges [EMAIL PROTECTED]:
Hadassa Brunschwig wrote:
Hi R-Help!
I am trying to use R2WinBUGS but I get the following error message in
WinBUGS
(and there must be something wrong with my R statement as I tried it
directly in
WinBUGS and it worked):
display(log)
check(C:/Documents and Settings/Daikon/Roche/pop_model.txt)
model is syntactically correct
data(C:/Documents and Settings/Daikon/Roche/data.txt)
expected key word structure
compile(7)
...(and of course nothing works after that)
and when I close WinBUGS i get:
Error in file(file, r) : unable to open connection
In addition: Warning message:
cannot open file 'codaIndex.txt'
Does anyone know what this 'expected key word structure' means?
This is my R code (and I guess my model file is ok):
modelA - c(C:/Documents and Settings/Daikon/Roche/pop_model.txt)
n - length(unique(subsetA$subject))#number of subjects
nt - 13 #number of days
Y - subsetA$concentr #concentration per
day/subject
t - 1:13 #days
dataA - list(n,nt,Y,t)
parameters -
c(tau,C0,st90,C0.pop,st90.pop,tau.cpop,tau.stpop)
inits -
list(tau=c(1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1),C0=5,st90=4,C0.pop=5,st90.pop=4,tau.cpop=0.2,tau.stpop=1)
mcmcA -
bugs(dataA,inits,parameters,modelA,debug=T,n.chains=7,bugs.directory=c:/Program
Files/WinBUGS14,working.directory=C:/Documents and
Settings/Daikon/Roche)
Which versions of R, WinBUGS and R2WinBUGS?
Your example is not reproducible for us.
I'd take a look whether dimensions are OK and whether subsetA$concentr
is in appropriate object, but without data and model file I am unable to
help for the data part.
For the inits part, please see ?bugs:
inits: a list with n.chains elements; each element of the list is itself
a list of starting values for the WinBUGS model, or a function creating
(possibly random) initial values. Alternatively, if inits = NULL,
initial values are generated by WinBUGS
Looks like you want to have the same inits for each chain. In order not
to repeat the inits 7 times, you might want to specify them simply as a
function such as:
inits - function(){
list(tau = rep(1, 17), C0 = 5, st90 = 4, C0.pop = 5, st90.pop = 4,
tau.cpop = 0.2, tau.stpop = 1)
}
Uwe Ligges
[Further correspondence on this particular topic, please respond to to
the package maintainer (Sibylle) and me directly rather than to R-help.]
Thanks so much...
Hadassa
__
R-help@stat.math.ethz.ch mailing list
https://stat.ethz.ch/mailman/listinfo/r-help
PLEASE do read the posting guide! http://www.R-project.org/posting-guide.html
