Statalist


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

Re: st: re: Is There a User-Written Command for Boosted Regression


From   Kit Baum <baum@bc.edu>
To   statalist@hsphsun2.harvard.edu
Subject   Re: st: re: Is There a User-Written Command for Boosted Regression
Date   Sun, 31 May 2009 13:24:30 -0400

<>
Sergiy said

2Kit: I think it depends on the perspective and the opposite
statement: "This is an illustration of why Macs are rather a pain" is
equally true. I can hardly think of any program that someone ported
from Mac to Win because there was no equivalent software in Windows,
while the reverse stream is endless. (I don't take QuickTime or iTunes
into account, as it is a proprietary file format limitation, not the
real demand force).

2David: If it is Intel based - you should be able to install Windows
on it, perhaps as a secondary OS to keep your current installation.
Having Intel CPU does not help much, because even the simpliest plugin
will probably request some resources (e.g. memory) at runtime from OS,
and it expects that OS understands the requests. So unless MacOs can
mimic Windows in every aspect of this sort (like some Linux
installations do, as I am told) there will remain an incompatibility.
So unless you really need smooth 256x256 icons and white keys on the
keyboard, perhaps it's better to stay with something more mainstream.

Microsoft claims that .Net products will work on Win and Mac machines
as long as an appropriate .Net-Framework is installed. (It is
Microsoft's response to the cross-platform portability) Unfortunatelly
Stata does not support it (while SPSS does):
http://www.spss.com/PDFs/pe.pdf


Whether software is ported from Win to Mac or from Mac to Win has very little to do with users' currentt platform of choice. The StataCorp philosophy is thankfully much broader in perspective than Sergiy's: the notion that Stata works the same way on all operating systems that are supported is certainly unique in the commercial statistical software arena, and it allows you to make use of the best tools for the job. For instance, our university has invested big bucks in a Linux cluster that has much more RAM and disk available than individual users, and Stata/MP is available there. But Windows- specific plugins are not. The managers of the cluster would laugh if someone suggested building an equivalent high-powered Windows cluster. "better to stay with something more mainstream" is the philosophy that has led US automakers GM and Chrysler to ruin, promoting mediocrity in design, customer satisfaction and durability. Those who believe that it would be much safer to stick with something more mainstream, like Windows Vista (about which I hear many complaints in the media) are barking up the same tree.

From the standpoint of making software available to others via the SSC Archive, it is a royal pain for a programmer to write a C plugin and produce and compile separate versions of it that will run on Windows-32, Windows-64, Mac OS X, Linux, AIX, and Solaris, because no one but StataCorp is likely to have easy access to all of those systems. So no one does. Development in Mata IMHO is a much better way to go, as there is little that Mata cannot do (relative to C/C++) and the performance hit is not too severe. If it was, I doubt that StataCorp would be so fond of doing all their development work in Mata nowadays.

Kit

PS> My colleagues in computational biology would argue strenouusly that they care very little about smooth 256x256 icons, white keys but need a stable, non-virus-ridden system whose OS does not eat up a significant fraction of their CPU power. Some choose Mac OS X, some choose Linux, but no one in that field that I know chooses Windows (well, maybe for Solitaire or Quake).

Kit Baum   |   Boston College Economics & DIW Berlin   |   http://ideas.repec.org/e/pba1.html
An Introduction to Stata Programming | http://www.stata-press.com/books/isp.html
   An Introduction to Modern Econometrics Using Stata  |   http://www.stata-press.com/books/imeus.html



*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/statalist/faq
*   http://www.ats.ucla.edu/stat/stata/



© Copyright 1996–2014 StataCorp LP   |   Terms of use   |   Privacy   |   Contact us   |   What's new   |   Site index