Roland <roland.rashleigh-berry@VIRGIN.NET> wrote [in part]:
> I've been working solidly learning Visual Basic (actually VB5 since I
> already have the learning edition and the CD Rom tutorial is
> far I believe it will indeed be possible to replace SAS with VB (so
> it runs under the right operating system), using Access as the native
For the most part. You could even write it in Python or Perl, using,
MySQL as the database, and have it run *everywhere*. For free. With no
run-time to sweat over.
> I think rewriting the stats procedures will be a doddle so
> the formulae are known (which they must be).
Supposedly. But, in fact, most of what I want SAS for is not a 'simple
formula'. You can write the formula down for the least-squares solution
a mixed model, but it is in matrix algebra and requires decent matrix
routines. Which VB is still lacking. [At least Perl now has a decent -
fast - matrix module, even if I have not been able to get it compiled on
boxen.] Or iterative solution systems. Or proper estimation of points
statistical distributions, with or without non-centrality paramaters.
Well, you get my drift.
> And these could be tested
> validated among the pharmaceutical companies and distributed free
Now *that* would be good. Validated components like that would be
> I would say that given the will to do it and the time then
> could be done.
I agree. But those are fairly large givens.
David Cassell, CSC
Senior computing specialist