R help - Dec 2001 - distributions w. skewness & kurtosis

Is there some reasonable way to generate random data from a
distribution that has some degree of skewness and/or kurtosis, but
would otherwise be normal?

thanks,
-------------- next part --------------
A non-text attachment was scrubbed...
Name: greiff.vcf
Type: text/x-vcard
Size: 398 bytes
Desc: Card for Warren R. Greiff
Url :
stat.ethz.ch/pipermail/r-help/attachments/20011210/2170bc6f/greiff.vcf

A similar question came up recently on the list (you can search the
archives for keywords like "kurtosis" at r-project.org, or more
easily
at Jonathan Baron's site, finzi.psych.upenn.edu/search.html).

r-project.org/nocvs/mail/r-help/2001/5320.html

  This query was about distributions with zero skewness and positive
kurtosis, but I think some of the answers may be helpful if you follow
them up.  Just to caution you, it's not clear what "would otherwise be
normal" means in this context -- I suppose you could say you want a
parametric family of distributions that includes the normal distribution
when some parameters are set to special values, but allow other
possibilities with non-zero higher cumulants ... (the gamma, t, and
possibly weibull distributions spring to mind).

r-project.org/nocvs/mail/r-help/2001/5320.html


On Mon, 10 Dec 2001, Warren R. Greiff wrote:
> Is there some reasonable way to generate random data from a
> distribution that has some degree of skewness and/or kurtosis, but
> would otherwise be normal?
>
> thanks,
-- 
318 Carr Hall                                bolker at zoo.ufl.edu
Zoology Department, University of Florida    zoo.ufl.edu/bolker
Box 118525                                   (ph)  352-392-5697
Gainesville, FL 32611-8525                   (fax) 352-392-3704

-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-
r-help mailing list -- Read ci.tuwien.ac.at/~hornik/R/R-FAQ.html
Send "info", "help", or "[un]subscribe"
(in the "body", not the subject !)  To: r-help-request at
stat.math.ethz.ch
_._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._

Many people pointed out that it wasn't clear what I meant by
"otherwise be normal?". To be frank, I wasn't quite sure what I
meant
myself, which was why I wrote "some reasonable way".  One condition I
had in mind was that if the parameters for skewness, and kurtosis were
set to 0, I'd be left with the normal distributions.  

Seems to me a number of responses may point me to solutions for my
problem.  Bob Wheeler's suggestion to look at the Johnson family of
distributions in SuppDists looks particularly promising, so I'll start
there.

The generalized lambda approach mentioned by Michaell Taylor looks
more complicated, but I found R code for it in the gld Package and
will look into that possibility too, when I get a chance.

thanks all,
-------------- next part --------------
A non-text attachment was scrubbed...
Name: greiff.vcf
Type: text/x-vcard
Size: 398 bytes
Desc: Card for Warren R. Greiff
Url :
stat.ethz.ch/pipermail/r-help/attachments/20011211/df273c9e/greiff.vcf