Re: stats

Moriyoshi, I forgot to say that examples most probably will be available soon written by Ed Luschei. He uses GD for visualisation. Andrey ----- Original Message ----- From: "Moriyoshi Koizumi" <moriyoshi@at.wakwak.com> To: "Andrey Hristov" <andrey@php.net> Cc: <internals@lists.php.net> Sent: Monday, August 11, 2003 10:08 PM Subject: Re: [PHP-DEV] [PECL-PROPOSAL] stats > As I can imagine some examples that demonstrate dynamic graph drawing with > ming or gd, I think it'd be cool to have. > > However, I looked at dcdflib.c and only found it was coded not in C but in > pseudo Fortran. That's worth a glance :) So the last concern is its > stability, since it simply uses static variables in the application-level > context (not in a thread-local context). This is fixable though. > > Moriyoshi > > "Andrey Hristov" <andrey@php.net> wrote: > > > Hi, > > few months ago I developed a wrapper around 2 scientific libraries. > > DCDFLIB (Library of C routines for Cumulative Distributions Functions, > > Inverses, and Other parameters)(by Barry Brown and James Lavato) > > and RANDLIB (by Barry Brown, James Lavato and Kathy Russell). > > I was asked to create the wrapper by a Mr. Ed Luschei from University > > of Wisconsin - Madison WI. He is a weed ecologist. The site of the > > "Weed ecology" group is : > > http://weedecology.wisc.edu/ . The wrapper is available > > here : > > http://www.hristov.com/andrey/projects/php_stuff/stats/ > > and can be downloaded from this url : > > > > http://www.hristov.com/andrey/projects/php_stuff/stats/stats.tgz > > There is previous an effort to wrap the library by 2 scientist of UCLA : > > Arno Ouwehand and Jan de Leeuw . Their implementation however is worse > > since they do not validate the input parameters. In their code there > > are available numerous problems that may lead to segmentation fault. > > Thus I decided to rewrite the wrapper and use different function names > > according to PHP's coding convention for function naming. > > The documentation of the 2 UCLA's scientist can be used as reference : > > http://www.jstatsoft.org/v02/i01/statistics.pdf . > > I propose this wrapper to be added to PECL. > > > Any comments/suggestions are welcome! > > > > Thanks, > > Andrey Hristov >