Varying-width strips give a compact illustration of a distribution.
The width of the strip is proportional to the density. This function
adds a varying-width strip to an exising plot.
Either the vector of points at which the density is
evaluated (if dens supplied), or a sample from the distribution
(if dens not supplied).
dens
Density at x. If dens is not supplied,
the density of the sample x is estimated by kernel density
estimation, using density(x,...).
at
Position of the centre of the strip on the y-axis (if
horiz=TRUE) or the x-axis (if horiz=FALSE).
width
Thickness of the strip at the maximum density, that is, the length of its
shorter dimension. Defaults to 1/20 of the axis range.
horiz
Draw the strip horizontally (TRUE) or vertically (FALSE).
scale
Alternative way of specifying the thickness of the
strip, as a proportion of width.
limits
Vector of minimum and maximum values, respectively, at which to terminate the strip.
col
Colour to shade the strip, either as a built-in R
colour name (one of colors()) or an RGB hex
value, e.g. black is "#000000".
border
Colour of the border, see polygon. Use
border=NA to show no border. The default, 'NULL', means to
use 'par("fg")' or its lattice equivalent
lwd
Line width of the border (defaults to
par("lwd") or its lattice equivalent).
lty
Line type of the border (defaults to
par("lty") or its lattice equivalent).
ticks
Vector of x-positions on the strip to draw tick
marks, or NULL for no ticks.
tlen
Length of the ticks, relative to the thickness of the strip.
twd
Line width of these marks (defaults to
par("lwd") or its lattice equivalent).
tty
Line type of these marks (defaults to
par("lty") or its lattice equivalent).
lattice
Set this to TRUE to make vwstrip
a lattice panel function instead of a base graphics function. panel.vwstrip(x,...) is equivalent to
vwstrip(x, lattice=TRUE, ...).
...
Additional arguments supplied to density(x,...), if
the density is being estimated.
Details
Varying-width strips look like violin plots. The difference is that
violin plots are intended to summarise data, while
vwstrip is
intended to illustrate a distribution arising from parameter
estimation or prediction. Either the distribution is known
analytically, or an arbitrarily large sample from the distribution is
assumed to be available via a method such as MCMC or bootstrapping.
Illustrating outliers is important for summarising data, therefore
violin plots terminate at the sample minimum and maximum and superimpose
a box plot (which appears like the bridge of a violin, hence the name).
Varying-width strips, however, are used to illustrate known
distributions which may have unbounded support. Therefore it is
important to think about where the strips should terminate (the
limits argument). For example, the end points may illustrate
a particular pair of extreme quantiles of the distribution.
The function vioplot in the vioplot
package and panel.violin in the lattice
package can be used to draw violin plots of observed data.
Author(s)
Christopher Jackson <chris.jackson@mrc-bsu.cam.ac.uk>
Hintze, J.L. and Nelson, R.D. (1998) Violin plots: a box plot -
density trace synergism. The American Statistician52(2),181–184.
See Also
denstrip, bpstrip, cistrip.
Examples
x <- seq(-4, 4, length=10000)
dens <- dnorm(x)
plot(x, xlim=c(-5, 5), ylim=c(-5, 5), xlab="x", ylab="x", type="n")
vwstrip(x, dens, at=1, ticks=qnorm(c(0.025, 0.25,0.5, 0.75, 0.975)))
## Terminate the strip at specific outer quantiles
vwstrip(x, dens, at=2, limits=qnorm(c(0.025, 0.975)))
vwstrip(x, dens, at=3, limits=qnorm(c(0.005, 0.995)))
## Compare with density strip
denstrip(x, dens, at=0)
## Estimate the density from a large sample
x <- rnorm(10000)
vwstrip(x, at=4)