http://www.lsoft.com/images/listserv-community.gifhttp://www.lsoft.com/images/listserv_small.gifBUGS ListBUGS List Archiveshttps://www.jiscmail.ac.uk/cgi-bin/webadmin?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/webadmin?A0=BUGS.

LISTSERV Web Interface 16.52019-11-13T08:41:36ZLawson, Andrew B.2019-11-12T20:35:20+00:002019-11-12T20:35:20+00:00new Bayesian Spatial Health modeling courses: March 2020, MUSC Charleston, SC, USAhttps://www.jiscmail.ac.uk/cgi-bin/webadmin?A2=BUGS;4b283110.1911Some places remain for a multi-day BDM Course Sequence for 2020<br><br>Introduction to Bayesian Disease Mapping (IBDM)<br>Bayesian Disease Mapping with INLA (BDMI)<br>Advanced Bayesian Disease Mapping (ABDM)<br><br>**Offering of Introductory and Advanced BDM courses and BDM with INLA in MUSC, Charleston, SC, USA**<br><br>Historic Charleston, South Carolina<br><br>*Course content*<br><br>These courses are designed to provide a comprehensive introduction to the area of<br>Bayesian disease mapping in applications to Public Health and Epidemiology:<br>The IBDM course will run on March 9th - 10th, the BDMI course will run on March 11th<br>and the ABDM course will run on March 12th [...] Shide Rafati2019-11-05T10:24:14+00:002019-11-05T10:24:14+00:00How to solve the error "Sorry something went wrong in procedure Stack. Value in module GraphStack"?https://www.jiscmail.ac.uk/cgi-bin/webadmin?A2=BUGS;eb5d026f.1911I run the below code to obtain posterior summaries using OpenBUGS. model is syntactically correct, dataloaded, model compiled, initial values generated and model initialized. But,while the model is updating I encounter the error:****** Sorry something wentwrong in procedure Stack. Value in module GraphStack ******<br><br>can anyone help me to solvethis error? <br><br>thanks.<br>################modelmodel { for (i in 1:252) {Sstar[i] <- p[i]+(1-p[i])*S0[i]; fstar[i] <- (1-p[i])*f0[i]S0[i] <- 1/(1+lam[i]*pow(t[i],kap))log(f0[i]) <- log(kap)+log(lam[i])+(kap-1)*log(t[i])-2*log(1+lam[i]*pow(t[i],kap))log(lam[i]) <-beta[1]+beta[2]*sex[i]+beta[3]*smoking[i]+beta[4]*Diabetes[i]+beta[5]*hypertension[i]+beta[6]*blood.a[i]+beta[7]*blood.b[i]+beta[8]*blood.ab[i]+beta[9]*marital_r[i]+beta[10]*job.house[i]+beta[11]*job.farm[i]+beta[12]*job.oth[i]+beta[13]*age[i]+beta[14]*diagnosis.age[i]+beta[15]*bmi[i]+beta[16]*hour.w[i]+beta[17]*time.w[i]+beta[18]*ch.k[i]+beta[19]*job.less[i]+beta[20]*job.retired[i]<br>p[i] <- 1/(1+exp(-(eta[1]+eta[2]*sex[i]+eta[3]*smoking[i]+eta[4]*Diabetes[i]+eta[5]*hypertension[i]+eta[6]*blood.a[i]+eta[7]*blood.b[i]+eta[8]*blood.ab[i]+eta[9]*marital_r[i]+eta[10]*job.house[i]+eta[11]*job.farm[i]+eta[12]*job.oth[i]+eta[13]*age[i]+eta[14]*diagnosis.age[i]+eta[15]*bmi[i]+eta[16]*hour.w[i]+eta[17]*time.w[i]+eta[18]*ch.k[i]+eta[19]*job.less[i]+eta[20]*job.retired[i])))# log-likelihood#log(L[i]) <- d[i]*log(fstar[i])+(1-d[i])*log(Sstar[i]);logL[i] < - log(L[i])zeros[i] < - 0zeros[i] ~ dloglik(logL[i])}Dv< - -2*sum(logL[])# priorseta[20]~dunif(-10, 10);eta[19]~dunif(-10, 10);eta[18]~dunif(-10, 10)eta[17]~dunif(-10, 10);eta[16]~dunif(-10, 10);eta[15]~dunif(-10, 10)eta[14]~dunif(-10, 10)beta[20]~dunif(-10, 10);beta[19]~dunif(-10, 10);beta[18]~dunif(-10, 10)beta[17]~dunif(-10, 10);beta[16]~dunif(-10, 10);beta[15]~dunif(-10, 10)beta[14]~dunif(-10, 10)for (j in [...]Phil Woodward2019-11-04T19:27:41-00:002019-11-04T19:27:41-00:00BugsXLA Ver.7 (A GUI for OpenBUGS or JAGS)https://www.jiscmail.ac.uk/cgi-bin/webadmin?A2=BUGS;6881fb5a.1911Hi,<br><br>BugsXLA (version 7) is a free to download GUI for OpenBUGS (WinBUGS) or<br>JAGS. It allows a wide range of GLMMs and other complex models to be fit to<br>data stored in Excel. BugsXLA uses model statements that should be familiar<br>to users of R, SAS or Genstat. Although the program avoids the need to code<br>up in the BUGS language, the files can be saved and rerun from within R or<br>edited directly by experienced users. [...] 小吉米2019-11-01T23:47:57+08:002019-11-01T23:47:57+08:00Fw: parameters' problemshttps://www.jiscmail.ac.uk/cgi-bin/webadmin?A2=BUGS;85b72864.1911| | 小吉米 邮箱：13240738478@163.com | 签名由 网易邮箱大师 定制 --------- Forwarded Message --------- From: 小吉米 Date: 11/01/2019 To: Subject: parameters' problems Hi everyone; I have used Winbugs for about 2 months, and I used it mainly for parameters’ estimation which also can be done by classical statistical MLE. I found somtimes the bayesian estimation and prediction suiting well with the MLE‘s results, but somtimes not. When reparalising models’ parameters with centering on the data's mean, the Markov chins converge so quickly, while not when directly using data. What i want is using the original data without centered on data's mean value, [...] Oliver Hooker2019-10-24T01:20:12+01:002019-10-24T01:20:12+01:002 Bayesian statistics courses with free accommodationhttps://www.jiscmail.ac.uk/cgi-bin/webadmin?A2=BUGS;748560b4.1910Applied Bayesian modelling for ecologists and epidemiologists (ABME05) FREE ACCOMMODATION AVAILABLE<br><br>https://www.prstatistics.com/course/applied-bayesian-modelling-for-ecologists-and-epidemiologists-abme05/<br><br>This course will be delivered by Dr. Mat Denwood from November 11th-15th in Glasgow City Centre<br><br>Course Overview:<br>This application-driven course will provide a founding in the basic theory & practice of Bayesian statistics, with a focus on MCMC modeling for ecological & epidemiological problems. Starting from a refresher on probability & likelihood, the course will take students all the way to cutting-edge applications such as state-space population modelling & spatial point-process modelling. By the end of the week, you should have a basic understanding of how common [...]Lawson, Andrew B.2019-10-13T15:27:45+00:002019-10-13T15:27:45+00:00new Bayesian Spatial Health modeling courses: March 2020, MUSC Charleston, SC, USAhttps://www.jiscmail.ac.uk/cgi-bin/webadmin?A2=BUGS;694d9c9b.1910Announcement: a multi-day Course Sequence for 2020<br><br>Introduction to Bayesian Disease Mapping (IBDM) Bayesian Disease Mapping with INLA (BDMI) Advanced Bayesian Disease Mapping (ABDM)<br><br>**Offering of Introductory and Advanced BDM courses and BDM with INLA in MUSC, Charleston, SC, USA**<br><br>Historic Charleston, South Carolina<br><br>*Course content*<br><br>These courses are designed to provide a comprehensive introduction to the area of Bayesian disease mapping in applications to Public Health and Epidemiology:<br>The IBDM course will run on March 9th - 10th, the BDMI course will run on March 11th and the ABDM course will run on March 12th - 13th 2020. [...] Goudie, Robert2019-09-10T13:17:37+00:002019-09-10T13:17:37+00:00COURSE: Bayesian statistics, MCMC and BUGS - MRC Biostatistics Unit, University of Cambridgehttps://www.jiscmail.ac.uk/cgi-bin/webadmin?A2=BUGS;fc185648.1909SHORT COURSE: Introduction to Bayesian statistics using BUGS<br><br>Weds 23rd – Thurs 24th October 2019 (2 days)<br>MRC Biostatistics Unit, University of Cambridge, UK<br><br>Course instructors: Dr Robert Goudie, Dr Anne Presanis, and Dr Chris Jackson (MRC Biostatistics Unit)<br><br>This course is intended to provide an introduction to Bayesian analysis and MCMC methods, and a fairly detailed tutorial on the use of OpenBUGS/JAGS/WinBUGS. [...]Samba Kumar2019-08-29T12:14:11+05:302019-08-29T12:14:11+05:30WinBUGS model implementation in NIMBLEhttps://www.jiscmail.ac.uk/cgi-bin/webadmin?A2=BUGS;61a3427b.1908Hi,<br><br>Have any of you in the list converted WinBUGS model to NIMBLE and<br>compared results.<br><br>I had implemented a hierarchical spatial (CAR) model with and without<br>indicator variables in WinBUGS and I am facing problems when I run the<br>same model in NIMBLE.<br><br>The posterior distributions of parameters for the model WITHOUT<br>indicator variables are very similar between WinBUGS and NIMBLE.<br>However, the model WITH indicator variables seems to have problem when<br>implemented in NIMBLE. The posterior summaries of indicator variables,<br>as well as the MC errors, obtained in NIMBLE are quite different from<br>WinBUGS estimates. I have used [...]Roy Levy2019-08-28T21:19:01-07:002019-08-28T21:19:01-07:003-Day Workshop: Introduction to Bayesian Statistical Modeling, Arizona State Universityhttps://www.jiscmail.ac.uk/cgi-bin/webadmin?A2=BUGS;e95d18a4.1908The Data in the Desert Training Workshop series is pleased to announce an<br>upcoming workshop:<br><br>Introduction to Bayesian Statistical modeling<br><br>Dates:<br>January 6-8, 2020 (Monday, Tuesday, Wednesday)<br><br>Location:<br>Tempe, Arizona<br><br>Instructor:<br>Dr. Roy Levy, Arizona State University<br>(roy.levy@asu.edu)<br><br>This three-day workshop assumes no prior experience with Bayesian<br>statistical modeling, and is intended as both a theoretical and practical<br>introduction. An understanding of Bayesian statistical modeling will be<br>developed by relating it to participants’ existing knowledge of traditional<br>frequentist approaches. To introduce Bayesian principles in familiar<br>contexts we will begin with simple binomial and univariate normal models,<br>then move to simple [...]Eun2019-08-28T10:59:35-04:002019-08-28T10:59:35-04:00Buttons such as density and stats do not get darkerhttps://www.jiscmail.ac.uk/cgi-bin/webadmin?A2=BUGS;3d777b43.1908I am using WinBUGS to to find the posterior distributions of the developed<br>PPoF models uncertain parameters. The code and data are listed below. I ran<br>the code. Model is synthetically correct. The data were loaded. I compiled.<br>Inits were generated. I could set c, n, s, n as nodes. When I enter * for<br>node, buttons such as density, stats, auto cor do not get darker so I<br>cannot press them. How can I fix it? Can anyone possibly help me out? Much<br>appreciated! [...]MILLARD, ANDREW R.2019-08-28T14:53:09+00:002019-08-28T14:53:09+00:00Re: Monitor could not be sethttps://www.jiscmail.ac.uk/cgi-bin/webadmin?A2=BUGS;ef5eb178.1908Under data you include<br>list(c=2, n=3, m=1, s=1, e=1)<br>if you load these nodes as data then you cannot monitor them.<br><br>Best wishes<br>AndrewEun2019-08-28T10:29:48-04:002019-08-28T10:29:48-04:00Monitor could not be sethttps://www.jiscmail.ac.uk/cgi-bin/webadmin?A2=BUGS;e030d92b.1908I am using WinBUGS to to find the posterior distributions of the developed<br>PPoF models uncertain parameters. The code and data are listed below. I ran<br>the code. Model is synthetically correct. The data were loaded. I compiled.<br>Inits were generated. When I tried to set c, n, s, n as nodes, I got an<br>error message that says "Monitor could not be set". How can I fix it? Can<br>anyone possibly help me out? Much appreciated! [...]Mohammad Abaei2019-08-23T08:36:59+00:002019-08-23T08:36:59+00:00Running Problemhttps://www.jiscmail.ac.uk/cgi-bin/webadmin?A2=BUGS;1c5138c7.1908Hi there,<br><br>I hope this email finds you well. Last week I come up with a problem with my code for compiling. The code synthetically is correct and the data can be loaded and initialized. However, when I want to update the model and get the result I am receiving that : [...] Ngandjui Cynthia2019-08-16T02:08:06+02:002019-08-16T02:08:06+02:00HELPhttps://www.jiscmail.ac.uk/cgi-bin/webadmin?A2=BUGS;b5331b44.1908Dear,<br>I have a couple of questions concerning winbugs and I am really struggling.<br>My first question is, I have a dataset with reaction time being the<br>response and I would like to obtain the posterior probability that reaction<br>time is increased by 10 per days. I know i need to use a step function but<br>I keep getting an error message when I try monitoring the node. I am also<br>trying to obtain a posterior predictive distribution at days 0-9 and have<br>no idea how to do that. Could you please help me with this?<br>Below are the codes [...]Bob OHara2019-08-12T11:47:30+02:002019-08-12T11:47:30+02:00Re: Error message in jags - Attempt to redefine nodehttps://www.jiscmail.ac.uk/cgi-bin/webadmin?A2=BUGS;2eb96c8e.1908(1) BUGS is not JAGS. The JAGS help forum is here:<br>https://sourceforge.net/p/mcmc-jags/discussion/610037/<br><br>(2) On line 4 you define y[t,x[t-1,m],m] as a dcat(), then on line 5 you<br>re-define the same nodes as x[t,m]. Poor JAGS doesn't know if they are<br>meant to be stochastic or deterministic.<br><br>Bob<br><br>On 8/12/19 3:46 AM, Istoni da Luz Sant'Ana wrote:<br>> 0<br>><br>><br>> I'm trying to run a non-homogeneous Markov Chain in JAGS. I have something wrong because when I run my code I get the following error:<br>><br>> RUNTIME ERROR: Compilation error on line 5. Attempt to redefine node y[2,2,1]<br> [...]Istoni da Luz Sant'Ana2019-08-12T02:46:00+01:002019-08-12T02:46:00+01:00Error message in jags - Attempt to redefine nodehttps://www.jiscmail.ac.uk/cgi-bin/webadmin?A2=BUGS;1f74732.19080<br><br>I'm trying to run a non-homogeneous Markov Chain in JAGS. I have something wrong because when I run my code I get the following error:<br><br>RUNTIME ERROR: Compilation error on line 5. Attempt to redefine node y[2,2,1]<br><br>I'd appreciate any help.<br><br>model<br>{<br>for (t in 2:T){<br>for (m in 1:M){<br>y[t,x[t-1,m],m]~dcat(p[t,m,x[t-1,m],1:J])<br>#Individual transition data where m represents individual index<br>y[t,x[t-1,m],m]<-x[t,m]<br>}} [...]alireza amoozmand2019-08-09T00:55:23+02:002019-08-09T00:55:23+02:00winbugs issuehttps://www.jiscmail.ac.uk/cgi-bin/webadmin?A2=BUGS;3cbdaf05.1908Hello,<br>I have two issues in winbugs and really apprciate your giudance.<br>I have a linear regression model and need to find the posterior probability<br>when lets say outcome in increased by 5 units.<br><br>the second question is obtaining posterior predictive distribution. When i<br>try what you proposed on stackexchange it fails to initialize. (model<br>compiles and data loads but fail to initialize) [...]SYMON K Ndung,u2019-07-16T15:49:26+01:002019-07-16T15:49:26+01:00specifying a new distribution in winbugshttps://www.jiscmail.ac.uk/cgi-bin/webadmin?A2=BUGS;28c6c216.1907Hello.<br>Am trying to code the Poisson Inverse Gaussian model which is a new distribution not included in the standard distribution. Requesting for assistance in implementing the model using the zeros trick. Kindly help with a sample code describing how to input its likelihood allowing for the mean and dispersion parameters to be modeled using glms or other non-linear approaches. [...]MILLARD, ANDREW R.2019-07-15T09:42:33+00:002019-07-15T09:42:33+00:00Re: Algorithms with discrete likelihood and continuous priorhttps://www.jiscmail.ac.uk/cgi-bin/webadmin?A2=BUGS;2cd51202.1907Compile your model and then use Node Info on the Info menu and it will tell you the updater being used. In this case Slice.Interval.<br><br>alpha does not have a Poisson likelihood, as it has a more complex relationship to the data p[i]. It is lambda that has a Poisson likelihood. [...]Leo Gürtler2019-07-12T18:38:45+02:002019-07-12T18:38:45+02:00exact Fisher testhttps://www.jiscmail.ac.uk/cgi-bin/webadmin?A2=BUGS;7216c46f.1907Dear all,<br><br>I wanted to reproduce the exact Fisher test as outlined in the BUGS book p.126 with both margins fixed and the dhyper() function.<br><br>Obviously something's wrong and I have not understood how to combine ps, m, and psi.<br><br>If both margins are fixed, I assume:<br><br>- p[2] = 1-p[1]<br>- psi is a consequence of both p's<br>- N is the sum of the single column n's [...] Ke Qin2019-07-11T18:32:59-04:002019-07-11T18:32:59-04:00Algorithms with discrete likelihood and continuous priorhttps://www.jiscmail.ac.uk/cgi-bin/webadmin?A2=BUGS;1b1c8d62.1907Dear Bugs community,<br><br>Hi there, I am trying to understand the following codes:<br><br>model {<br>for (i in 1:N) {<br>p[i] ~ dpois(lambda)<br>}<br>lambda <- 2*exp(-2*alpha*3)/(2*pow(4,2))<br>alpha ~ dnorm(beta,tau)T(0,0.2)<br>beta ~ dnorm(0,10000)<br>tau ~ dgamma(2,0.01)<br>}"<br><br>for alpha, it has poisson likelihood and normal distribution. I am<br>wondering what algorithms used here to run MCMC. ARS? Metropolis Hasting?<br>or Slice Sampling?<br><br>I wrote some slice sampling codes in R and the results are slightly<br>different (-0.79 vs -0.83). I am wondering if my codes are wrong or slice<br>sampling is not appropriate here. [...]Auke Tas2019-07-02T10:40:11+01:002019-07-02T10:40:11+01:00Initializing hierarchical model with missing valueshttps://www.jiscmail.ac.uk/cgi-bin/webadmin?A2=BUGS;43b84ea5.1907Deal all,<br><br>I have some issues with my model. I am trying to model three dependent variable using a logit specification with hierarchical priors on the coefficients. For some reason I keep running into either the error 'postcondition violated' or 'undefined real result'. What I could find online was that either the prior or the initial values is likely to be the problem and after playing around with them I managed to come up with a model that works so far, however there are two thing that are still puzzling me: [...]