Date: Mon, 15 Jan 2007 00:12:54 -0800
Reply-To: David L Cassell <davidlcassell@MSN.COM>
Sender: "SAS(r) Discussion" <SAS-L@LISTSERV.UGA.EDU>
From: David L Cassell <davidlcassell@MSN.COM>
Subject: Re: A new statistical programming language
In-Reply-To: <CA8F89971ADA9F47A6C915BA2397844201EA7654@MAILBE2.westat.com>
Content-Type: text/plain; format=flowed
HERMANS1@WESTAT.COM replied:
>
>Robert:
>Thanks for the advice. I admit that I find clinical trial programming
>standards in the pharmaceutical industry, well, amusing. Despite that I
>have lasted longer than most. Perhaps I have managed to get by because I
>don't feel compelled to put too much weight on weak premises and
>extraordinary claims.
>
>I encourage programmers to challenge use of inferior methods, whether
>endorsed by a lead statistician or not. You chose arbitrary substitution of
>value associated with a maximum date as an imputation method. While I can
>replicate the result you present in your example, I would not know how to
>implement this arbitrary method outside the scope of your example. How do
>visits acquire values of -1, for instance?
>
>The SQL query that I wrote conforms to ANSI SQL syntax. A new
>implementation and extension of the SQL language might give you a better
>starting point.
>
>A number of talented programmers have developed new public-use programming
>languages, C, C++, awk, perl, python, ruby, and xlisp, for example, and
>have not encountered the same hostile resistance that you have. I would
>blame lethargy, not hostile resistance to change, for lack of enthusiasm
>for your new statistical programming language. Is anyone gaining an
>advantage over fellow programmers by using it? Most of the single source
>languages begin gaining popularity among a small group of devoted users,
>then gather momentum.
>
>I have to agree with Toby. The advantages that you are touting for your new
>programming language don't inspire much enthusiasm. An example by someone
>other than the developer that shows how your language solves his or her
>programming problem would have more impact than whining about not being
>recognized due to the hostility of the statistical programming
>establishment (on SAS-L, two guys from the Pacific NW).
>
>The name that you have given the language may also be a problem. Those
>O'Reilly books have great drawings of animals on their covers. I don't see
>what they could do for Vilno. Maybe a potrait of Pushkin....
>Siguru
The open-source languages that Sig mentioned have - on occasion -
already generated substantial stat and math programming aid. So
maybe that's a way to go here. XLisp spawned XLispStat, which
is nice and LISP-y too. Perl spawned a *ton* of math and stat
modules, including the PDL module which is pretty darn fast.
there are a bunch of math/stat modules for C/C++/C#/Java, and
there are a decent number for programs like Python, Ruby, VB,
etc. I don't think it would make much sense to extend awk to
serious stat problems, but that doesn't mean someone hasn't tried...
Now Robert has been working on Vilno for most of a decade, as
I understand it, so he has a good understanding of how long it
might take to get serious statistical computing integrated in.
R has a ton of excellent people contributing code, and their
stuff for mixed models is still not up to snuff after all this time.
This stuff takes a lot of person-hours.
David
--
David L. Cassell
mathematical statistician
Design Pathways
3115 NW Norwood Pl.
Corvallis OR 97330
_________________________________________________________________
Type your favorite song. Get a customized station. Try MSN Radio powered
by Pandora. http://radio.msn.com/?icid=T002MSN03A07001
| 
