[R] CircStats reveals underlying R bug?
bolker at zoo.ufl.edu
Wed Nov 12 14:28:25 CET 2003
Maybe you should take this up with package maintainers (who may or may
not be reading R-help) ... this sounds like a design/documentation issue
rather than a "bug" per se (although the distinction is not always clear).
To be honest, the underlying R code in CircStats doesn't seem terribly
sophisticated -- I have hacked it some for my own use (e.g. to allow
the various von Mises functions (rvm, dvm, etc.) to handle vectors of
different parameter values).
On Wed, 12 Nov 2003, John Christie wrote:
> On Nov 12, 2003, at 7:04 AM, Peter Dalgaard wrote:
> > John Christie <jc at or.psychology.dal.ca> writes:
> >> circ.plot in CircStats under R 1.8.0 yields incorrect plots (easily
> >> compare hist() and rose.diag()). However, the code in circ.plot has
> >> not changed. Has anyone tracked this down yet?
> > Could you supply a directly runnable example, please?
> Actually, in making something concise and runnable I found the problem.
> It is undocumented that circ.plot does not handle negative angles
> while rose.diag does. I think that should be called a bug in
> circ,plot. It seems to me the code of these two should be merged
> anyway to avoid future errors. rose.diag can do a circ.plot type
> diagram (but you always get the rose as well).
> So, a simple example is.
> circ.plot(x, bins=18, stack=T, shrink=2)
> rose.diag(x, bins=18, pts=T, shrink=2)
> # everything should look OK up until you try comparing the next two
> circ.plot(-x, bins=18, stack=T, shrink=2)
> rose.diag(-x, bins=18, pts=T, shrink=2)
> R-help at stat.math.ethz.ch mailing list
620B Bartram Hall bolker at zoo.ufl.edu
Zoology Department, University of Florida http://www.zoo.ufl.edu/bolker
Box 118525 (ph) 352-392-5697
Gainesville, FL 32611-8525 (fax) 352-392-3704
More information about the R-help