tRophicPosition (version 0.8.0)

multiModelTP: Multiple model calculation of trophic position

Description

This function takes an isotopeData class object and calculates by default three Bayesian models: one and two baselines without carbon fractionation and two baselines with carbon fractionation.

This function takes an isotopeData class object and calculates by default three Bayesian models: one and two baselines without carbon fractionation and two baselines with carbon fractionation.

Usage

multiModelTP(
  siData = siData,
  lambda = 2,
  n.chains = 2,
  n.adapt = 20000,
  n.iter = 20000,
  burnin = 20000,
  thin = 10,
  models = c("oneBaseline", "twoBaselines", "twoBaselinesFull"),
  params = NULL,
  print = FALSE,
  quiet = FALSE,
  ...
)
multiModelTP(
  siData = siData,
  lambda = 2,
  n.chains = 2,
  n.adapt = 20000,
  n.iter = 20000,
  burnin = 20000,
  thin = 10,
  models = c("oneBaseline", "twoBaselines", "twoBaselinesFull"),
  params = NULL,
  print = FALSE,
  quiet = FALSE,
  ...
)

Value

For each model calculated, returns a data frame of 4 elements with raw posterior samples, a list with posterior TP samples, a list with posterior muDeltaN (if one baseline model was chosen) or alpha (if a two baselines model was chosen) and a data frame with a summary of posterior samples named gg.

For each model calculated, returns a data frame of 4 elements with raw posterior samples, a list with posterior TP samples, a list with posterior muDeltaN (if one baseline model was chosen) or alpha (if a two baselines model was chosen) and a data frame with a summary of posterior samples named gg.

Arguments

siData

an isotopeData class object.

lambda

numerical value, represents the trophic level of baseline(s).

n.chains

number of parallel chains for the model. If convergence diagnostics (such as Gelman-Rubin) are printed, n.chains needs to be >= 2.

n.adapt

number of adaptive iterations, before the actual sampling.

n.iter

number of iterations for Bayesian modelling (posterior sampling).

burnin

number of iterations discarded as burn in.

thin

thinning. Number of samples discarded while performing posterior sampling.

models

string or list representing Bayesian models. At the moment they can be "oneBaseline", "twoBaselines" and/or "twoBaselinesFull".

params

aditional parameters included as a list.

print

logical value to indicate whether Gelman and Rubin's convergence diagnostic and summary of samples are printed.

quiet

logical value to indicate whether messages generated during compilation will be suppressed, as well as the progress bar during adaptation.

...

additional arguments passed to this function.

Examples

if (FALSE) {
isotopeData <- generateTPData()
models <- multiModelTP(isotopeData, n.adapt = 500, n.iter = 500,
burnin = 500)
credibilityIntervals(models$gg, x = "model")
}
if (FALSE) {
isotopeData <- generateTPData()
models <- multiModelTP(isotopeData, n.adapt = 500, n.iter = 500,
burnin = 500)
credibilityIntervals(models$gg, x = "model")
}