In the moment I am writing an R function for drawing bagplots (two dimensional boxplots). For some elements of the plot are found numerically the resulting plots differs a little bit from the correct bagplots. Here is the link to the actual version: http://www.wiwi.uni-bielefeld.de/~wolf/software/R-wtools/bagplot/bagplot.R ( R code ) http://www.wiwi.uni-bielefeld.de/~wolf/software/R-wtools/bagplot/bagplot.pdf (examples and description) Now I am looking for examples to test / to improve my function. Where can I get data sets and the corresponding correct bagplots? Thanks Peter Wolf (pwolf at wiwi.uni-bielefeld.de)
bagplot - a rough R function
3 messages · Martin Maechler, Peter Wolf
"Peter" == Peter Wolf <pwolf at wiwi.uni-bielefeld.de>
on Wed, 21 Sep 2005 16:49:10 +0200 writes:
Peter> In the moment I am writing an R function for drawing
Peter> bagplots (two dimensional boxplots).
Peter> For some elements of the plot are found numerically
Peter> the resulting plots differs a little bit from the correct bagplots.
Peter> Here is the link to the actual version:
Peter> http://www.wiwi.uni-bielefeld.de/~wolf/software/R-wtools/bagplot/bagplot.R
Peter> ( R code )
Peter> http://www.wiwi.uni-bielefeld.de/~wolf/software/R-wtools/bagplot/bagplot.pdf
Peter> (examples and description)
Peter> Now I am looking for examples to test / to improve my function.
Peter> Where can I get data sets and the corresponding correct bagplots?
I'll send you my (unpublished) version of my package 'bagplot'
{{the one I've never released because of segmentation fault bugs
somewhere in Fortran}} in a private E-mail.
In my version I had started make a function
bag()
which *computes* all the information needed and then returns an
(S3) object "bag". Then, the plotting happens via
the
plot.bag <- function(.....) {..........}
method.
So you can say
plot(bag(.........), ........)
if you want both (computing and plotting) at once, but can also
separate the two --- following very much the modern "S" way
"whole object ..".
Martin
Martin Maechler wrote:
"Peter" == Peter Wolf <pwolf at wiwi.uni-bielefeld.de>
on Wed, 21 Sep 2005 16:49:10 +0200 writes:
Peter> In the moment I am writing an R function for drawing Peter> bagplots (two dimensional boxplots). Peter> For some elements of the plot are found numerically Peter> the resulting plots differs a little bit from the correct bagplots. Peter> Here is the link to the actual version: Peter> http://www.wiwi.uni-bielefeld.de/~wolf/software/R-wtools/bagplot/bagplot.R Peter> ( R code ) Peter> http://www.wiwi.uni-bielefeld.de/~wolf/software/R-wtools/bagplot/bagplot.pdf Peter> (examples and description) Peter> Now I am looking for examples to test / to improve my function. Peter> Where can I get data sets and the corresponding correct bagplots? I'll send you my (unpublished) version of my package 'bagplot' {{the one I've never released because of segmentation fault bugs somewhere in Fortran}} in a private E-mail. In my version I had started make a function bag() which *computes* all the information needed and then returns an (S3) object "bag". Then, the plotting happens via the plot.bag <- function(.....) {..........} method. So you can say plot(bag(.........), ........) if you want both (computing and plotting) at once, but can also separate the two --- following very much the modern "S" way "whole object ..". Martin
Hello Martin, you are right. Separating computing and plotting is a better way for a public version and I think that I will split my function. But during testing algorithmic problems it is nice to have only one function. Peter