R vs. S-PLUS

Gary Cable writes:
 > I recently joined Insightful Corporation as Product Manager for Financial
 > Solutions. I continue to hear a theme of R versus S and I find it somewhat
 > curious. 

The theme that I hear most often is R versus S+, both being
implementations of the S language.

 > it can also be a weakness because it (IMO) does not represent a
 > commercial tool that provides me with stability, product support,
 > and professional services.

During 2001, I used both R and S+ simultaneously for some relatively
serious work. I found R superior then in terms of stability and
product support. Others opinions may differ.

 > R is primarily an academic solution. 

This is your opinion. Billions of dollars in real assets around the
globe are run using R every day. Maybe those managers are all idiots,
but I doubt it.

 > Insightful owns the IP (the S Language) and is committed to its ongoing
 > improvement.

Does Insightful own the S language? I do not believe that this is true.

Dave Kane

PS. My purpose here is not to start a flame war. I can think of
reasons why someone in finance might prefer S+ to R --- easy data
retrieval from Factset and Bloomberg comes to mind. But there was no
way that I could let a claim about R being "primarily an academic
solution" go unchallenged.

-----Original Message-----
From: r-sig-finance-bounces@stat.math.ethz.ch 
[mailto:r-sig-finance-bounces@stat.math.ethz.ch] On Behalf Of 
Gary Cable
Sent: Tuesday, November 30, 2004 9:58 AM
To: r-sig-finance@stat.math.ethz.ch
Subject: [R-sig-finance] R vs. S-PLUS

Hello all, 

I recently joined Insightful Corporation as Product Manager 
for Financial Solutions. I continue to hear a theme of R 
versus S and I find it somewhat curious. R is based on S and 
has gained a following based on a number of contributions 
from academia and industry experts.
 
I prefer to view R as one evolutionary path that S has taken. 
It has the same positives that any other GNU tool that I have 
seen in my professional life--it is free and open for 
everyone's use and contribution. This is both it's strength 
and it's weakness. It is a strength because people can use it 
as a sandbox for their ideas; it can also be a weakness 
because it (IMO) does not represent a commercial tool that 
provides me with stability, product support, and professional 
services. 

Hi Gary,


Welcome to the exciting world of S.

I guess you are really new to Insightful. My more than a decade experience
is different from what you outlined above. I used to use S-Plus until about
year 2000 in a company with a decent IT budget. Most of the so called
support was coming from the S-news list. I did contact Insightful for
support on issues I could not resolve on the S-news list. In both cases the
answer was "it is too deep in the language, we can not fix it". One of those
issues was related to the introduction of the new S classes in S-Plus (they
were basically unusable at that time). A similar introduction in R has
recently gone very smoothly.
In 2000 or 2001 (R-1.3) I figured R to be a superior system to S-Plus and
had switched, some of my fellow colleagues followed the suit. I've found R
to be very stable as far as our production requirements are concerned. It
has a very predictable release schedule with a procedure for features
deprecation and removal so you are never caught by a surprise.

The support that comes from R-help list is more than enough and is 24*7*365.
The only real drawback is that the language of the replies is sometimes too
"mentoring" so sensitive people can get upset (this is where Insightful does
have an edge). Bugs get fixed almost instantly.

I am sure you know, but just in case, the "no warranty and no support"
clauses in GPL are merely to protect the software contributor against
lawsuits in some litigious environments, they are not representative of the
actual support you are going to have with the software.

Please let me know if there is anything else to product support that I
didn't address.


R is not perfect, with it I too have run into a couple of issues (related to
the speed of file reading and database connectivity) which I was not able to
resolve on R-help / R-devel list to my satisfaction. In both cases I looked
at the underlying C code, figured out the problem and wrote a couple of
packages to work around them. I had the very same problems with S-Plus too
and the only option I had back then was to use Perl for IO intensive jobs.

Now about the professional services. You can get R training, you can hire an
R consultant to write a specialized package for you, etc. I was surprised to
learn how much of a supply of such services is out there.

Sorry if you already knew all of this. I just wanted to give you some info
that you might find useful in your new role of the product manager.

Regards,
Vadim

I don't want to detract from the success of the R language, but if I were
going to build solutions in a highly regulated industry that expects
documentation, official support and standardization -- I would be inclined
toward a commercial solution rather than a freeware solution. This decision
comes from twenty years of experience of application development. 
 

The usual argument for preferring commercial software is that the commercial
software will be better written, tested, documented and supported.  The 
expectation
is that the free software will have many bugs, and the commercial 
software will
have few bugs.

Perhaps my experience is unusual, but in using S-PLUS for Windows for 4 
months
I found 2 serious bugs, 2 very annoying bugs and a few minor bugs.  In 
2.5 years
of intense use of R I have found 4 or 5 minor bugs (almost all of which 
have now
been fixed).

I don't think playing the "commercial" card for S-PLUS is a good 
strategy until
the point that it has at least as good of quality control as R.

Patrick Burns

Burns Statistics
patrick@burns-stat.com
+44 (0)20 8525 0696
http://www.burns-stat.com
(home of S Poetry and "A Guide for the Unwilling S User")