[R-pkgs] New package: ``denstrip'' for compactly illustrating distributions

Chris Jackson chris.jackson at mrc-bsu.cam.ac.uk
Wed Aug 27 15:38:57 CEST 2008


Dear R users,

I'd like to announce a new package on CRAN called ``denstrip''.  It
implements ``density strips'' and other graphical methods for
illustrating and comparing distributions in a compact fashion.

Posterior distributions of parameters are often summarised using point
and line drawings of means and credible intervals.  This is common,
for example, in multiple regression or meta-analysis.  Density
strips generalise these to illustrate whole distributions.  Instead of a
point and line, a shaded strip indicates the density as proportional to
the darkness of the shading.  They taper to white at the end of the
strip, instead of terminating at a clear limit - this may discourage
casually categorising effects as ``significant'' if the line excludes
the null.

The shading idea generalises to ``density regions'' to show
uncertainty about continuously-varying quantities, such as predictions
from time series.  The package includes other functions for
illustrating distributions in ``one dimension'', such as varying-width
strips (similar to violin plots) and sectioned density plots.

If you're interested in reading more about these methods, I discuss
them in a forthcoming article in The American Statistician, ``Displaying
uncertainty with shading'', also available from

http://www.mrc-bsu.cam.ac.uk/personal/chris/papers/denstrip.pdf

Comments and suggestions for improvement of the package are welcome.

-- 
Christopher Jackson <chris.jackson at mrc-bsu.cam.ac.uk>
Research Statistician, MRC Biostatistics Unit, Institute of Public
Health, Robinson Way, Cambridge, UK, CB2 0SR. +44 (1223) 330381




More information about the R-packages mailing list