[Top][All Lists]

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: Study about accuracy of statistical software, incl. Octave

From: Marco Caliari
Subject: Re: Study about accuracy of statistical software, incl. Octave
Date: Wed, 25 Mar 2009 08:58:03 +0100 (CET)
User-agent: Alpine 1.00 (DEB 882 2007-12-20)

On Tue, Mar 24, 2009 at 8:49 PM, Jason Riedy <address@hidden> wrote:
And Jaroslav Hajek writes:
But what is the correct answer? Just writing those numbers into Octave
introduces an error ~ 3e-17. I *don't* expect the "correct" answer is
1e7 + 0.2. You can't expect Octave to guess your actual numbers.

I do expect better than a single precision result from std(), however.
I'll try to poke at what happens (esp. using DNRM2) this week, although
I'd love for someone to beat me to it...


I would be interested in creating improved compiled versions of mean,
meansq, std and var. Any other good pointers for algorithms than

Chan, Tony F.; Golub, Gene H.; LeVeque, Randall J. (1983). Algorithms for Computing the Sample Variance: Analysis and Recommendations. The American Statistician 37, 242-247.

Ling, Robert F. (1974). Comparison of Several Algorithms for Computing Sample Means and Variances. Journal of the American Statistical Association, Vol. 69, No. 348, 859-866.


reply via email to

[Prev in Thread] Current Thread [Next in Thread]