From: Paul A Bristow (pbristow_at_[hidden])
Date: 2007-01-30 14:07:34
>[mailto:boost-bounces_at_[hidden]] On Behalf Of Eric Niebler
>Sent: 30 January 2007 18:35
>To: boost_at_[hidden]; Matthias Troyer
>Subject: Re: [boost] Accumulators review
>I'm including Matthias Troyer on this, in the hopes he might have some
>feedback about your kurtosis suggestion .
Some useful links are
and most definitely, as always,
Weisstein, Eric W. "Kurtosis." From MathWorld--A Wolfram Web Resource. http://mathworld.wolfram.com/Kurtosis.html
As one who has been rescused from this pit by John more than once,
I strongly recommend using kurtosis for the 'proper' version,
and always saying kurtosis excess (kurtosis - 3) whenever applicable.
We have eventually done this with the math Toolkit distribution properties.
Unwisely IMO, Wikipedia have not done this, and there is much confusion,
and some mistakes that I have corrected.
PS and about your torture test - I don't think this should be the focus of the review.
There are horses for courses, and the not-always-accurate version should be fine for most applications.
(If the data are going to be as unreasonable as the torture data are, one could argue that they are wrong!
and that there should have never have been used to calculate a simple mean etc anyway - it needs a robust mean?).
--- Paul A Bristow Prizet Farmhouse, Kendal, Cumbria UK LA8 8AB +44 1539561830 & SMS, Mobile +44 7714 330204 & SMS pbristow_at_[hidden]
Boost list run by bdawes at acm.org, gregod at cs.rpi.edu, cpdaniel at pacbell.net, john at johnmaddock.co.uk