Many users experience long execution times and convergence warnings when trying to fit complex linear mixed-effects models with lmer. I have, in the past, shown that such models can be fit using the MixedModels ( https://github.com/dmbates/MixedModels.jl) package for Julia ( https://julialang.org) and that the data can be pulled from an R representation using either the RCall ( https://github.com/JuliaInterop/RCall.jl) or RData ( https://github.com/JuliaData/RData.jl). Recently the JuliaCall package for R ( https://github.com/Non-Contradiction/JuliaCall) has become available on CRAN. I have a short note at http://rpubs.com/dmbates/377897 on how to use that package to fit models using MixedModels from R.
Vignette on using Julia MixedModels from R
8 messages · Douglas Bates, HAJDUK Gabriela, Rolf Turner +2 more
On 13/04/18 03:33, Douglas Bates wrote:
Many users experience long execution times and convergence warnings when trying to fit complex linear mixed-effects models with lmer. I have, in the past, shown that such models can be fit using the MixedModels ( https://github.com/dmbates/MixedModels.jl) package for Julia ( https://julialang.org) and that the data can be pulled from an R representation using either the RCall ( https://github.com/JuliaInterop/RCall.jl) or RData ( https://github.com/JuliaData/RData.jl). Recently the JuliaCall package for R ( https://github.com/Non-Contradiction/JuliaCall) has become available on CRAN. I have a short note at http://rpubs.com/dmbates/377897 on how to use that package to fit models using MixedModels from R.
This is of considerable interest to me since I am involved in some consulting work (about which I have annoyed the r-sig-mixed-models list on previous occasions!) which involves mixed models and I have experienced the problems referred to in the opening line of your message. So I had a look at http://rpubs.com/dmbates/377897 just now, and it looks promising. I have however a possibly naive or misguided question: How do I get a printed copy of the vignette? I tried saving or printing the page that showed up in my browser (Firefox; GoogleChrome) in various ways. The result always seems to get truncated at a single page, with the bulk of the document omitted. (The document is also somewhat messed up in other ways.) Is there a simple way of saving a copy of the vignette in such a way that I can print off a hard copy? I suppose I am simply demonstrating my ignorance and showing that I am out of the loop when it comes to dealing with Rpubs and Rmarkdown documents, but that's the way it is. Thanks for any tips and advice. cheers, Rolf
Technical Editor ANZJS Department of Statistics University of Auckland Phone: +64-9-373-7599 ext. 88276
Hello Rolf, You can highlight all text (Ctrl/Cmd + A) and then right click and choose print. It should then give you the entire highlighted content - I?m on Chrome and if I just right click and print I get the first page only, but highlighting beforehand gives all the content. You might need to play with scaling before printing. Best, Gabriela
On 13 Apr 2018, at 02:11, Rolf Turner <r.turner at auckland.ac.nz> wrote: On 13/04/18 03:33, Douglas Bates wrote:
Many users experience long execution times and convergence warnings when trying to fit complex linear mixed-effects models with lmer. I have, in the past, shown that such models can be fit using the MixedModels ( https://github.com/dmbates/MixedModels.jl) package for Julia ( https://julialang.org) and that the data can be pulled from an R representation using either the RCall ( https://github.com/JuliaInterop/RCall.jl) or RData ( https://github.com/JuliaData/RData.jl). Recently the JuliaCall package for R ( https://github.com/Non-Contradiction/JuliaCall) has become available on CRAN. I have a short note at http://rpubs.com/dmbates/377897 on how to use that package to fit models using MixedModels from R.
This is of considerable interest to me since I am involved in some consulting work (about which I have annoyed the r-sig-mixed-models list on previous occasions!) which involves mixed models and I have experienced the problems referred to in the opening line of your message. So I had a look at http://rpubs.com/dmbates/377897 just now, and it looks promising. I have however a possibly naive or misguided question: How do I get a printed copy of the vignette? I tried saving or printing the page that showed up in my browser (Firefox; GoogleChrome) in various ways. The result always seems to get truncated at a single page, with the bulk of the document omitted. (The document is also somewhat messed up in other ways.) Is there a simple way of saving a copy of the vignette in such a way that I can print off a hard copy? I suppose I am simply demonstrating my ignorance and showing that I am out of the loop when it comes to dealing with Rpubs and Rmarkdown documents, but that's the way it is. Thanks for any tips and advice. cheers, Rolf -- Technical Editor ANZJS Department of Statistics University of Auckland Phone: +64-9-373-7599 ext. 88276
_______________________________________________ R-sig-mixed-models at r-project.org mailing list https://stat.ethz.ch/mailman/listinfo/r-sig-mixed-models
The University of Edinburgh is a charitable body, registered in Scotland, with registration number SC005336.
On 13/04/18 21:10, HAJDUK Gabriela wrote:
Hello Rolf, You can highlight all text (Ctrl/Cmd + A) and then right click and choose print. It should then give you the entire highlighted content - I?m on Chrome and if I just right click and print I get the first page only, but highlighting beforehand gives all the content. You might need to play with scaling before printing.
Thanks very much Gabriela. It works like a charm with Chrome, though not (at all) with Firefox. cheers, Rolf
On 13 Apr 2018, at 02:11, Rolf Turner <r.turner at auckland.ac.nz> wrote: On 13/04/18 03:33, Douglas Bates wrote:
Many users experience long execution times and convergence warnings when trying to fit complex linear mixed-effects models with lmer. I have, in the past, shown that such models can be fit using the MixedModels ( https://github.com/dmbates/MixedModels.jl) package for Julia ( https://julialang.org) and that the data can be pulled from an R representation using either the RCall ( https://github.com/JuliaInterop/RCall.jl) or RData ( https://github.com/JuliaData/RData.jl). Recently the JuliaCall package for R ( https://github.com/Non-Contradiction/JuliaCall) has become available on CRAN. I have a short note at http://rpubs.com/dmbates/377897 on how to use that package to fit models using MixedModels from R.
This is of considerable interest to me since I am involved in some consulting work (about which I have annoyed the r-sig-mixed-models list on previous occasions!) which involves mixed models and I have experienced the problems referred to in the opening line of your message. So I had a look at http://rpubs.com/dmbates/377897 just now, and it looks promising. I have however a possibly naive or misguided question: How do I get a printed copy of the vignette? I tried saving or printing the page that showed up in my browser (Firefox; GoogleChrome) in various ways. The result always seems to get truncated at a single page, with the bulk of the document omitted. (The document is also somewhat messed up in other ways.) Is there a simple way of saving a copy of the vignette in such a way that I can print off a hard copy? I suppose I am simply demonstrating my ignorance and showing that I am out of the loop when it comes to dealing with Rpubs and Rmarkdown documents, but that's the way it is. Thanks for any tips and advice.
On 04/13/18 21:58, Rolf Turner wrote:
On 13/04/18 21:10, HAJDUK Gabriela wrote:
Hello Rolf, You can highlight all text (Ctrl/Cmd + A) and then right click and choose print.
It should then give you the entire highlighted content - I'm on Chrome and if I just right click and print I get the first page only, but highlighting beforehand gives all the content. You might need to play with scaling before printing. Thanks very much Gabriela. It works like a charm with Chrome, though not (at all) with Firefox.
In Firefox (at least version 61+, but probably earlier), right click, then select "This frame", then "Show only this frame", then print.
Jonathan Baron, Professor of Psychology, University of Pennsylvania Home page: http://www.sas.upenn.edu/~baron
On 13/04/18 22:42, Jon Baron wrote:
On 04/13/18 21:58, Rolf Turner wrote:
On 13/04/18 21:10, HAJDUK Gabriela wrote:
Hello Rolf, You can highlight all text (Ctrl/Cmd + A) and then right click and choose print.
It should then give you the entire highlighted content - I'm on Chrome and if I just right click and print I get the first page only, but highlighting beforehand gives all the content. You might need to play with scaling before printing. Thanks very much Gabriela.? It works like a charm with Chrome, though not (at all) with Firefox.
In Firefox (at least version 61+, but probably earlier), right click, then select "This frame", then "Show only this frame", then print.
Yep, that works. Thanks. Not exactly intuitive, but, is it? (The same must be said for Gabriela's solution.) Software designers often seem to ignore the needs of "ordinary" users. :-( cheers, Rolf
Technical Editor ANZJS Department of Statistics University of Auckland Phone: +64-9-373-7599 ext. 88276
Works nicely but I?m trying to track down how to move data back and forth. I need to use glmm for predictions.
On Fri, Apr 13, 2018 at 5:49 PM Rolf Turner <r.turner at auckland.ac.nz> wrote:
On 13/04/18 22:42, Jon Baron wrote:
On 04/13/18 21:58, Rolf Turner wrote:
On 13/04/18 21:10, HAJDUK Gabriela wrote:
Hello Rolf, You can highlight all text (Ctrl/Cmd + A) and then right click and choose print.
It should then give you the entire highlighted content - I'm on Chrome and if I just right click and print I get the first page only, but highlighting beforehand gives all the content. You might need to play with scaling before printing. Thanks very much Gabriela. It works like a charm with Chrome, though not (at all) with Firefox.
In Firefox (at least version 61+, but probably earlier), right click, then select "This frame", then "Show only this frame", then print.
Yep, that works. Thanks. Not exactly intuitive, but, is it? (The same must be said for Gabriela's solution.) Software designers often seem to ignore the needs of "ordinary" users. :-( cheers, Rolf -- Technical Editor ANZJS Department of Statistics University of Auckland Phone: +64-9-373-7599 ext. 88276
_______________________________________________ R-sig-mixed-models at r-project.org mailing list https://stat.ethz.ch/mailman/listinfo/r-sig-mixed-models
Sent from Gmail Mobile [[alternative HTML version deleted]]
On 14/04/18 09:59, Mitchell Maltenfort wrote:
Works nicely but I?m trying to track down how to move data back and forth. I need to use glmm for predictions.
Does not work nicely for me. I tried (as per Doug's instructions):
> system.time(j <- julia_setup(JULIA_HOME=jh,verbose=TRUE))
where "jh" had been assigned the character string
"/home/rolf/Desktop/Julia/julia-d386e40c17/bin/"
which is where I have the Julia binary currently living.
I got the response:
> Julia version 0.6.2 at location
/home/rolf/Desktop/Julia/julia-d386e40c17/bin will be used.
> Julia initiation...
> Finish Julia initiation.
> Loading setup script for JuliaCall...
> Segmentation fault (core dumped)
I have contacted the maintainer of the JuliaCall package, and he has
responded and is looking into the problem. (I am running R from the
command line under Ubuntu 16.04.)
Has anyone else encountered such a seg fault?
cheers,
Rolf
Technical Editor ANZJS Department of Statistics University of Auckland Phone: +64-9-373-7599 ext. 88276