Skip to contents

Extract and export Mplus trace and autocorrelation plots into pdf

Source: R/MplusHelpers.R
traceplots_mplus.Rd

The function exports all the available plots for each parameter and every chain. Therefore it can take a long time and the resulting pdf file can be large. Each page in pdf is for a specific parameter.

Usage

traceplots_mplus(
  x,
  is.file = F,
  pdffile = NULL,
  gg = T,
  param.id = NULL,
  raster = F,
  niter = NULL,
  chains = NULL,
  PSR = TRUE,
  PSR.version = "Rstan",
  ESS = T
)

Arguments

x

Either a model read by readModels, OR a gh5 file produced by Mplus (in Mplus it should be stated PLOTS: TYPE IS PLOT2;)

is.file

Logical, if x argument is file name. Default is FALSE, (i.e. it's an R object, a model read by readModels)

pdffile

File name of the pdf report to export all the plots. Defaults to an of x argument with a '.pdf' extension when is.file == TRUE, and extracts the original filename of the out file if x is a readModels object.

gg

Logical. Whether ggplot or base R graphics should be used. Default is ggplot which is slower but more flexible, e.g. with many chains.

param.id

Single interger or a vector. Single intterger shows 1:param.id parameters to plot, a vector shows parameters by their order id. If NULL (default), all the parameters are used.

raster

Logical. Currently not used. It was an option to convert vector plots to raster plots to save space and produce smaller pdfs. Currently not used.

niter

How many iterations should be shown on a traceplot. NULL (default) refers to all iterations. However, traceplots for long chains may result in highly overplotted graphics which are slow and take a lot of space on disk. A specified number will take a random sample of values from the iterations.

PSR

If PSR (R-hat) should be printed for each parameter. See eachParamPSRMplus.

PSR.version

Version of PSR; default is "Rstan" - the most up-to-date, provided by the Rstan::Rhat function (Vehtari et al., 2019). Other possible values: "Gelman" (Gelman et al. (2004)); or "Mplus" (Asparouhov and Muthen, 2010), and "Naive" which is a raw ratio of sum of a between- and within-chain variances to a within-chain variance.

ESS

Logical. If effective sample size should be printed on the plots. See ess_bulk

Examples


if (FALSE) { # \dontrun{
m <- MplusAutomation::readModels("mybayes.out")
convergencePlotsMplus(m, niter = 5000, PSR.version = "Mplus")

# or directly from gh5 file:
convergencePlotsMplus("mybayes.gh5", param.id = 1, PSR.version = "Rstan", gg=F)
} # }