rmsb: stanDxplot – R documentation

Become an expert in R — Interactive courses, Cheat Sheets, certificates and more!

stanDxplot

Diagnostic Trace Plots

Description

For an rms Bayesian fit object, uses by default the stored posterior draws to check convergence properties of posterior sampling. If instead rstan=TRUE, calls the rstan traceplot function on the rstan object inside the rmsb object, to check properties of posterior sampling. If rstan=TRUE and the rstan object has been removed and previous=TRUE, attempts to find an already existing plot created by a previous run of the knitr chunk, assuming it was the plotno numbered plot of the chunk.

Usage

stanDxplot(
  x,
  which = NULL,
  rstan = FALSE,
  previous = TRUE,
  plotno = 1,
  rev = FALSE,
  stripsize = 8,
  ...
)

Arguments

`x`	an `rms` Bayesian fit object
`which`	names of parameters to plot, defaulting to all non-intercepts. When `rstan=FALSE` these are the friendly `rms` names, otherwise they are the `rstan` parameter names. If the model fit was run through `stackMI` for multiple imputation, the number of traces is multiplied by the number of imputations. Set to `'ALL'` to plot all parameters.
`rstan`	set to `TRUE` to use `rstan::traceplot()` on a (presumed) stored `rstan` object in `x`, otherwise only real iterations are plotted and parameter values are shown as points instead of lines, with chains separated
`previous`	see details
`plotno`	see details
`rev`	set to `TRUE` to reverse direction for faceting chains
`stripsize`	specifies size of chain facet label text, default is 8
`...`	passed to `rstan::traceplot()`

Value

ggplot2 object if rstan object was in x

Author(s)

Frank Harrell

rmsb

Bayesian Regression Modeling Strategies

v0.0.2

GPL (>= 3)

Authors

Frank Harrell [aut, cre] (<https://orcid.org/0000-0002-8271-5493>), Ben Goodrich [ctb] (contributed Stan code), Ben Bolker [ctb] (wrote original code that is folded into the pdensityContour function), Doug Bates [ctb] (write original code for highest posterior density interval that is folded into the HPDint function)

Initial release

2021-02-27