search

brianstock / MixSIAR

A framework for Bayesian mixing models in R:
http://brianstock.github.io/MixSIAR/
94 stars 75 forks source link

Problem with "large" dataset using ID as fixed effect #159

Open kscharnweber opened 5 years ago

kscharnweber commented 5 years ago

Hi MixSIAR folks!

I am a big fan of your package and have used it a lot for my research!

I have discovered one issue when working with MixSIAR, and I was wondering if you have heard about it before and weather you can help me with it. With my dataset, I am interested in getting individual diet contribution obtained from d13C and d15N for fish and compare them with individual fatty acid data. Therefore, I am using fish ID as a fixed factor and apply the Process only error structure. I have a dataset of more than 70 individuals, and here is the problem: The model runs without any problem and indicates that it is done. I then would like to receive the output and the diagnostics file. What the package does it to produce one posterior plot for each individual. However, it will stop after individual 62 and R gives me the error message that there are too many open devices. So I will never receive the summary statistics and diagnostics file.

What can I do? Is there any R command that I can use directly in the R console to the receive the two files? Let me know if I should provide a mock data set to try it out.

Thanks so much for your help!

All the best, Kristin

goldsoundz80 commented 5 years ago

Hey I'm not affiliated with MixSIAR but I had the same issue as you and figured out the solution. Plotting so many charts is causing the error. You can suppress the chart plotting by editing the following part of your R script. First find the output_options bit in the code for processing the JAGs output, and make sure sup post = TRUE

output_options <- list(summary_save = TRUE,
                       summary_name = "summary_statistics",
                       sup_post = TRUE,

I ran into another issue next, my summary statistics.txt cut off after a certain number of individuals. if you get this too, you'll notice at the end of your summary statistics that it says it reached the max.print limit.

You can show the currently set limit by: getOption("max.print")

You can set a new one with (you can change the number to whatever you like): options(max.print = 99999)

I just added the line to my MixSIAR script.

kscharnweber commented 5 years ago

Hi goldsounz80,

thank you so much for your reply! Seems as if this could be a solution for my problem. However, what is adding some complication is that I am, so far, using the MixSIAR GUI, so the code is running somewhere in the backend and I do not really see it and have no chance to modify it. The GUI actually produces a text file with the MixSIAR code, but this is before the output (no output_options line) and if I let it run in the console, things do not really work. What would be the command to obtain the exact code of my JAGS model that I ran using the GUI, including the faulty output_options line? If I could recieve that, I could easily modify it and run it using the R console instead of the GUI, right?

All the best, Kristin