http://www.lsoft.com/images/listserv-community.gifhttp://www.lsoft.com/images/listserv_small.gifBUGS ListBUGS List Archiveshttps://www.jiscmail.ac.uk/cgi-bin/wa-jisc.exe?RSS&L=BUGS&v=ATOM1.0
This is an Atom formatted XML site feed. It is intended to be viewed in a Newsreader.
Alternatively you can view the HTML archives at https://www.jiscmail.ac.uk/cgi-bin/wa-jisc.exe?A0=BUGS.
LISTSERV Web Interface 16.52024-01-17T13:06:59ZBob O'Hara2024-01-17T14:01:34+01:002024-01-17T14:01:34+01:00Re: Initial valueshttps://www.jiscmail.ac.uk/cgi-bin/wa-jisc.exe?A2=BUGS;af9ab533.2401No, BUGS needs the actual values (either ecause you give them, or<br>because you tell it to generate them). I'm afraid it's not enough to<br>just specify their prior (which is what you have done).<br><br>Bob<br><br>On 17/01/2024 13:55, José María Grisolía Santos wrote:<br>> Thanks a lot Bob.<br>> However, I told OpenBUGS that<br>> brwqs[i]~dnorm(mrwqs,srwqs) And then provide mean and SD. Isn't it<br>> enough for initial values?<br>><br>> Thank you<br>><br>><br>> Enviado desde Outlook para iOS <https://aka.ms/o0ukef><br>> ------------------------------------------------------------------------<br>> *De:* Bob O'Hara <bob.ohara@ntnu.no><br>> *Enviado:* Wednesday, January 17, 2024 12:36:49 PM<br>> *Para:* José [...]Bob O'Hara2024-01-17T13:36:49+01:002024-01-17T13:36:49+01:00Re: Initial valueshttps://www.jiscmail.ac.uk/cgi-bin/wa-jisc.exe?A2=BUGS;3d42e0dc.2401You don't initialise brwqs, which I'm guessing isn't in your data. You<br>might want to use a larger value of srwqs as an initial value, because<br>BUGS will draw inits for brwqs from a normal with the initial values<br>given by the inits for the hyper parameters, so unless the variance of<br>rwq is small, you will generate values a long way from the posterior,<br>and this might make convergence difficult. [...]José María Grisolía Santos2024-01-17T12:10:41+00:002024-01-17T12:10:41+00:00Initial valueshttps://www.jiscmail.ac.uk/cgi-bin/wa-jisc.exe?A2=BUGS;67b48869.2401Dear all<br>This is a mixed logit model prepared for individual parameter estimation in one of the parameters (brwqs[i]). The model compiles, though when I launch the initial parameters<br><br>list(bcte= 0.57639, mrwqs=0.0516, srwqs=0.001, bpin=-0.5,<br> bbbs=0.04, bbill=-0.334)<br><br>is telling me "initial values loaded but chain contains uninitialized variables"<br><br>But this is not true. All initial values are provided here. If I generate values it tricks another error. Can anyone spot the problem? [...]Sorif Hossain2024-01-15T16:58:40+06:002024-01-15T16:58:40+06:00Query about WinBUGS/OpenBUGS/MultiBUGS softwarehttps://www.jiscmail.ac.uk/cgi-bin/wa-jisc.exe?A2=BUGS;d0eb2240.2401USA_Counties.txt<br><https://drive.google.com/file/d/1lze6bKrulFCkWV4QkVMqxeEVNLyQPeNb/view?usp=drive_web><br>Dear BUGS admin,<br><br>I hope you are well. Recently I have worked with WInBUGS, OpenBUGS and<br>MUltiBUGS software to analyse spatial analysis using USA county data. The<br>county number was 3143 and the neighbour was 18000+. When I run the code it<br>shows an index out of range. Even I can not import the county data for a<br>large number of counties. Could you kindly help me in this case? My code<br>and data are ok but I think it's a software limitation. Could you kindly<br>help me in this case? Could you kindly import this Splus file [...]Pablo Verde2023-12-18T21:54:12+01:002023-12-18T21:54:12+01:00Re: something went wrong in procedure LN in math module beta regression modelhttps://www.jiscmail.ac.uk/cgi-bin/wa-jisc.exe?A2=BUGS;f510de8.2312Thanks, Berwin. Here is the corrected BUGS script:<br><br>model<br>{<br># Datos<br>for (i in 1:n) {<br><br>logit(mu[i]) <- beta0 + beta1 * explicativa[i]<br><br>a[i] <- mu[i]*phi<br>b[i] <- (1-mu[i])*phi<br><br>respuesta[i] ~ dbeta(a[i], b[i])<br>prediccion[i] ~ dbeta(a[i], b[i])<br># respuesta[i] ~ dbeta(1, mu[i]) # Distribución beta para la<br>variable respuesta<br>}<br><br>phi ~ dunif(0, 100) # phi>0 restriction<br><br># Priors<br>beta0 ~ dnorm(0, 0.001) # Prior para el intercepto<br>beta1 ~ dnorm(0, 0.001) # Prior para la pendiente<br>} [...]Pablo Verde2023-12-18T21:12:06+01:002023-12-18T21:12:06+01:00Re: something went wrong in procedure LN in math module beta regression modelhttps://www.jiscmail.ac.uk/cgi-bin/wa-jisc.exe?A2=BUGS;c31d1c3e.2312Hi Berwin,<br><br>Yes, this is correct!<br><br>E(y[i]) = mu[i]<br><br>and<br><br>Var(y[i]) = mu[i]*(1-mu[i])/(1+phi)<br><br>The model is parametrized as mean and precision.<br><br>Cheers,<br><br>Pablo<br><br>On 2023-12-18 15:55, Berwin A Turlach wrote:<br>> G'day Pablo,<br>><br>> On Mon, 18 Dec 2023 13:50:37 +0100<br>> Pablo Verde <PabloEmilio.Verde@HHU.DE> wrote:<br>><br>> Shouldn't this<br>><br>>> a[i] <- mu[i]*phi<br>>> b[i] <- mu[i]*(1-mu[i])*phi<br>><br>> be<br>><br>>> a[i] <- mu[i]*phi<br>>> b[i] <- (1-mu[i])*phi<br>><br>> So that dbeta(a[i], b[i]) has mean mu[i] and variance<br>> mu[i]*(1-mu[i])/(phi+1)?<br>><br>> Cheers,<br>><br>> Berwin Berwin A Turlach2023-12-18T22:55:22+08:002023-12-18T22:55:22+08:00Re: something went wrong in procedure LN in math module beta regression modelhttps://www.jiscmail.ac.uk/cgi-bin/wa-jisc.exe?A2=BUGS;92b93947.2312G'day Pablo,<br><br>On Mon, 18 Dec 2023 13:50:37 +0100<br>Pablo Verde <PabloEmilio.Verde@HHU.DE> wrote:<br><br>Shouldn't this<br><br>> a[i] <- mu[i]*phi<br>> b[i] <- mu[i]*(1-mu[i])*phi<br><br>be<br><br>> a[i] <- mu[i]*phi<br>> b[i] <- (1-mu[i])*phi<br><br>So that dbeta(a[i], b[i]) has mean mu[i] and variance<br>mu[i]*(1-mu[i])/(phi+1)?<br><br>Cheers,<br><br>Berwin Pablo Verde2023-12-18T13:50:37+01:002023-12-18T13:50:37+01:00Re: something went wrong in procedure LN in math module beta regression modelhttps://www.jiscmail.ac.uk/cgi-bin/wa-jisc.exe?A2=BUGS;76255c10.2312Nestor,<br><br>You have to parametrize the Beta regression according to<br>the parameter restrictions.<br><br>The following code works well:<br><br>model<br>{<br># Datos<br>for (i in 1:n) {<br><br>logit(mu[i]) <- beta0 + beta1 * explicativa[i]<br><br>a[i] <- mu[i]*phi<br>b[i] <- mu[i]*(1-mu[i])*phi<br><br>respuesta[i] ~ dbeta(a[i], b[i])<br>prediccion[i] ~ dbeta(a[i], b[i])<br># respuesta[i] ~ dbeta(1, mu[i]) # Distribución beta para la<br>variable respuesta<br>}<br><br>phi ~ dunif(0, 100) # phi>0 restriction [...]Néstor Javier Roncancio Duque2023-12-16T15:04:02-05:002023-12-16T15:04:02-05:00something went wrong in procedure LN in math module beta regression modelhttps://www.jiscmail.ac.uk/cgi-bin/wa-jisc.exe?A2=BUGS;d1255bc8.2312Regards<br><br>I need your help please, I am try to run this model:<br><br>model<br><br>{<br><br># Datos<br><br>for (i in 1:n) {<br><br>mu[i] <- beta0 + beta1 * explicativa[i]<br><br>respuesta[i] ~ dbeta(1, mu[i]) # Distribución beta para la variable<br>respuesta<br><br>}<br><br># Priors<br><br>beta0 ~ dnorm(0, 0.001) # Prior para el intercepto<br><br>beta1 ~ dnorm(0, 0.001) # Prior para la pendiente<br><br>}<br><br># Datos [...]Ashley Sanchez2023-08-18T11:23:56-05:002023-08-18T11:23:56-05:00Applied Bayesian Data Analysis Seminarshttps://www.jiscmail.ac.uk/cgi-bin/wa-jisc.exe?A2=BUGS;8a658ec.2308Hello everyone,<br><br>Statistical Horizons will host two seminars on Bayesian methods, both<br>taught by Roy Levy, this fall.<br><br>*Applied Bayesian Data Analysis<br><https://statisticalhorizons.com/seminars/applied-bayesian-data-analysis/?utm_source=referral&utm_medium=referral&utm_campaign=abd-fall23-listserv>,<br>September 7-9. *In this seminar, you will get both a practical and<br>theoretical introduction to Bayesian methods. Bayesian principles will be<br>introduced in familiar contexts starting with simple binomial and<br>univariate normal models and then moving to simple and multiple regression. [...]