Bug #66983 for Statistics-Test-WilcoxonRankSum: bug when changing exact

Sun Mar 27 11:10:17 2011 idit [...] nrgene.com - Ticket created

Subject:	bug when changing exact_upto
Date:	Sun, 27 Mar 2011 17:09:48 +0200
To:	bug-Statistics-Test-WilcoxonRankSum [...] rt.cpan.org
From:	Idit Buch <idit [...] nrgene.com>

Hello, When I increase "exact_upto" to just 25 (Statistics::Test::WilcoxonRankSum->new({ exact_upto => 25})) and load two samples of 12 instances each, then the program gets stuck. Please advise thanks Idit Buch, PhD Computational Scientist Nrgene Ltd.

Sun Mar 27 16:13:42 2011 INGRIF [...] cpan.org - Correspondence added

Could you post your data? It seems to work for me - it just takes a very long time if you want to compute the exact test for N > 20. I tried with 48 items and had to kill the program after some minutes. Another comment: you don't need to set exact_upto, you can enforce exact calculations by calling the test like this: $wilcox_test->probability_exact(); (after loading the data) But as I sad, it can take a very long time. Best, Ingrid

Sun Mar 27 16:13:43 2011 The RT System itself - Status changed from 'new' to 'open'

Mon Mar 28 05:38:49 2011 idit [...] nrgene.com - Correspondence added

Subject:	Re: [rt.cpan.org #66983] bug when changing exact_upto
Date:	Mon, 28 Mar 2011 11:38:21 +0200
To:	bug-Statistics-Test-WilcoxonRankSum [...] rt.cpan.org
From:	Idit Buch <idit [...] nrgene.com>

Dear Ingrid, Thank you for your reply. My data contains numbers in range [0,1). Is there any intention to optimize the perl code so that it'll work much faster ? regards Idit Buch, PhD Computational Scientist Nrgene Ltd. On Sun, Mar 27, 2011 at 10:13 PM, Ingrid Falk via RT < bug-Statistics-Test-WilcoxonRankSum@rt.cpan.org> wrote: Show quoted text

> <URL: https://rt.cpan.org/Ticket/Display.html?id=66983 > > > Could you post your data? > > It seems to work for me - it just takes a very long time if you want to > compute the exact test for N > 20. I tried with 48 items and had to kill > the program after some minutes. > > Another comment: you don't need to set exact_upto, you can enforce exact > calculations by calling the test like this: > > $wilcox_test->probability_exact(); > > (after loading the data) > > But as I sad, it can take a very long time. > > Best, > > Ingrid >

Mon Mar 28 05:54:50 2011 ingrid.falk [...] loria.fr - Correspondence added

Subject:	Re: [rt.cpan.org #66983] bug when changing exact_upto
Date:	Mon, 28 Mar 2011 11:54:23 +0200
To:	bug-Statistics-Test-WilcoxonRankSum [...] rt.cpan.org
From:	Ingrid Falk <ingrid.falk [...] loria.fr>

On 03/28/2011 11:38 AM, Idit Buch via RT wrote: Show quoted text

> Queue: Statistics-Test-WilcoxonRankSum > Ticket <URL: http://rt.cpan.org/Ticket/Display.html?id=66983 > > > Dear Ingrid, > > Thank you for your reply. > My data contains numbers in range [0,1). > Is there any intention to optimize the perl code so that it'll work > much faster ?

No, I'm afraid not. It could be improved to some extent, but in fact the computational complexity is inherent to the problem, I mean the problem is such that it is bound to take a lot of computation time for larger N. Why don't you want to use the normal approximation, I thought that's what's usually recommended for N > 20? In the mean time I came across R - an interpreted language optimised for all kind of statistics problems. It is much better suited for these kind of computations than Perl and also has a wilcox.test routine: http://stat.ethz.ch/R-manual/R-patched/library/stats/html/wilcox.test.html R (http://www.r-project.org/) is open source and was easy to install and run (for me, on Linux/Ubuntu). Best, Ingrid

Message body is not shown because sender requested not to inline it.

Download signature.asc
application/pgp-signature 294b

Message body not shown because it is not plain text.

Mon Mar 28 07:25:55 2011 idit [...] nrgene.com - Correspondence added

Subject:	Re: [rt.cpan.org #66983] bug when changing exact_upto
Date:	Mon, 28 Mar 2011 13:25:27 +0200
To:	bug-Statistics-Test-WilcoxonRankSum [...] rt.cpan.org
From:	Idit Buch <idit [...] nrgene.com>

Thanks so much for your suggestions. Idit On Mon, Mar 28, 2011 at 11:54 AM, ingrid.falk@loria.fr via RT < bug-Statistics-Test-WilcoxonRankSum@rt.cpan.org> wrote: Show quoted text

> <URL: http://rt.cpan.org/Ticket/Display.html?id=66983 > > > On 03/28/2011 11:38 AM, Idit Buch via RT wrote:

> > Queue: Statistics-Test-WilcoxonRankSum > > Ticket <URL: http://rt.cpan.org/Ticket/Display.html?id=66983 > > > > > Dear Ingrid, > > > > Thank you for your reply. > > My data contains numbers in range [0,1). > > Is there any intention to optimize the perl code so that it'll work > > much faster ?

> > No, I'm afraid not. It could be improved to some extent, but in fact the > computational complexity is inherent to the problem, I mean the problem > is such that it is bound to take a lot of computation time for larger N. > > Why don't you want to use the normal approximation, I thought that's > what's usually recommended for N > 20? > > In the mean time I came across R - an interpreted language optimised for > all kind of statistics problems. It is much better suited for these kind > of computations than Perl and also has a wilcox.test routine: > > http://stat.ethz.ch/R-manual/R-patched/library/stats/html/wilcox.test.html > > R (http://www.r-project.org/) is open source and was easy to install and > run (for me, on Linux/Ubuntu). > > Best, > > Ingrid > > > > -----BEGIN PGP SIGNATURE----- > Version: GnuPG v1.4.10 (GNU/Linux) > Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org/ > > iF4EAREIAAYFAk2QWtwACgkQ5AS2AFBGpVhTyAD⣖鄂聯ᤃ薓䚣縤浫캝 > 3O464jYmRfDO2WPSdL8A/jAhO0kGWJKglx1VviZyaF1/Z7sxHlPEkAPI1kgaZ6AZ > =kA8/ > -----END PGP SIGNATURE----- > >

-- Idit Buch, PhD Computational Scientist Nrgene Ltd.

Bug #66983 for Statistics-Test-WilcoxonRankSum: bug when changing exact_upto