crossplot {tbea}

crossplot: Plot the median and HPD interval bars for pairs of distribution

Description

crossplot: Plot the median and HPD interval bars for pairs of distribution

Usage

crossplot(
  log1Path,
  log2Path,
  skip.char = "#",
  pattern = NULL,
  idx.cols = NULL,
  bar.lty,
  bar.lwd,
  identity.lty,
  identity.lwd,
  extra.space = 0.5,
  ...
)

Arguments

log1Path

character vector of length 1. Path to the first log file.

log2Path

character vector of length 1. Path to the second log file.

skip.char

character vector of length 1, with '#' as default value. Which symbol is used as a comment. This will allow to ignore lines which start with the symbol when reading data.

pattern

character vector of length 1. the pattern for subsetting the columns containing the data to be plotted.

idx.cols

either an integer vector with the position of the columns to pick, or a character vector with the column names to pick. Defaults to NULL.

bar.lty

The line type to be used as error bars.

bar.lwd

As above but the width

identity.lty

The line type to be used in the identity y = x line

identity.lwd

As above but the width.

extra.space

numeric vector of length 1. How much space to be allowed in both xlim and ylim depending on the smallest value in highest density intervals plus or minus extra.space. A value of 0.5 units on the dimension of interest is used by default.

...

Optional arguments to be passed to 'plot' such as 'main', 'xlab', 'ylab', 'pch' and 'cex'.

Details

The function produces a crossplot, which is a scatterplot where we are comparing two distributions associated to each point by means of the medians as the points, and the highest density intervals as bars around the point. For instance, x may represent the prior of a set of parameters while y represents the posterior. Error bars on the x axis then are highest density intervals from the prior, and those on the y axis represent the interval for the posterior.

This function can also be used to compare two independent runs for (visual) convergence: If they are sampling the same posterior distribution, then they should fall on the identity y=x line.

Value

This function returns nothing, it plots to the graphical device.

Author(s)

Gustavo A. Ballen

Examples

## Not run: 
crossplot(log1Path="log1.tsv", log2Path="log2.tsv", skip.char="#",
          pattern="par", cols=NULL, bar.lty=1, bar.lwd=1,
          identity.lty=2, identity.lwd=1,
          extra.space=0.5, main="My plot", xlab="log 1 (prior)", ylab="log 2 (posterior)", pch=19)

## End(Not run)