Extracting Posterior Distribution

[Bethke, Bethany (DNR)]

Hello!

 

I’m using tRophicPosition and wondering if it’s possible to get the posterior distribution estimates out of the model, for both alpha and TP (and for multiple species, if possible). Essentially I want the numbers that make up the panels on the right side of this plot:

Love the package – thanks so much for your work!

 

Bethany Bethke

Fisheries Research Scientist | Fish and Wildlife

 

[Claudio Quezada Romegialli]

Hi Bethany

It depends on how you got those plots. If you have posterior estimates for your variables, let say like this:

isotopeData <- "Your data for one species"

model.string <- jagsBayesianModel()
model <- TPmodel(data = isotopeData, model.string = model.string,
                 n.adapt = 5000)
posterior.samples <- posteriorTP(model = model, n.iter = 20000,
                                 variable.names = c("TP", "muDeltaN"))

summary(posterior.samples)

Then you have to combine the chains (by default TPmodel runs 2 chains in parallel):

posterior.combined <- coda::mcmc(do.call(rbind, posterior.samples))

posterior.combined is an mcmc object with the number of objects declared in posteriorTP (see variable.names above). So in this case, we have 2 objects, the first is TP, and the second is muDeltaN:

hist(posterior.combined[,1])
hist(posterior.combined[,2])

If you want the "raw posterior data" just use as.numeric:

as.numeric(posterior.combined[,1]) #for TP

as.numeric(posterior.combined[,2]) #for muDeltaN

When you have multiple species or multiple models, then you will have a dataframe with the results, in this case called "models":

isotopeData <- generateTPData()
models <- multiModelTP(isotopeData, n.adapt = 200, n.iter = 200,
                       burnin = 200)
credibilityIntervals(models$gg, x = "model")

models$TP is a list with the raw posterior TP data (a list)

models$alpha is a list with the raw posterior alpha data (a list)

models$gg is a dataframe with some summary results of the models run; and

models$samples are the raw data but as mcmc objects

I believe this is what you want. If you have any doubt, take a look at the vignettes attached.

Cheers

Claudio

--
To download tRophicPosition package visit https://github.com/clquezada/tRophicPosition
---
You received this message because you are subscribed to the Google Groups "tRophicPosition" group.
To unsubscribe from this group and stop receiving emails from it, send an email to trophicposition-s...@googlegroups.com.
To view this discussion on the web, visit https://groups.google.com/d/msgid/trophicposition-support/BN8PR09MB3572D4F1C06C04E152206983DE4A0%40BN8PR09MB3572.namprd09.prod.outlook.com.