Date: Fri, 13 Nov 2009 11:26:35 -0500
Reply-To: oloolo <dynamicpanel@YAHOO.COM>
Sender: "SAS(r) Discussion" <SAS-L@LISTSERV.UGA.EDU>
From: oloolo <dynamicpanel@YAHOO.COM>
Subject: Re: Another SAS v. R thread
I totally agree with you and tried to talk to SAS developers at regional
If SAS can't integrate IML(which I don't like at all) into SAS/BASE, at
least make the JAVA interface better.
On Thu, 12 Nov 2009 18:10:10 -0800, xlr82sas <xlr82sas@AOL.COM> wrote:
> I use both R and SAS. My goal is to combine the two so that the
>result is greater than either alone. Back in the late 80s 'proc
>matrix' was part of base SAS and it even had 128bit numerics(at least
>on mainframes), if only SAS had kept IML in SAS proper.
>The real power of R is the 2000+ R packages. The packages seem to be
>growing exponentially ove r the last year.
>Here are some weak points in R but not SAS
> 1. Until about a week ago I had not seen a nice implementation of
>associative arrays(hashes) in R. However I notice recentlly,
>11/4/2009, a HASH package for R on CRAN. http://cran.r-
> 2. No native SQL funcionality, although R provides interfaces to
>just about any database
> 3. Have not seen extensive native regular expression functions in R
>(maybe there is a package)
> 4. I don't believe R runs on IBM mainframs under zOS, OS/390,
>MVS., z/VM. (should work under LINUX)
> 5. I don't believe R has native functions for the many
>transformations that SAS can do. Things like date format, S390 format,
>PIB, Zoned decimal....
> 6. I have not seen the nice reporting functions in R like 'proc
>report', 'proc tabulate' , templates, tagsets and ods
> 7. I have had issues trying to get help when one of the packages did
>not work as promised.
> 8. For spliting, merging, updating, joining, keyed lookups(VSAM
>ISAM) of large files. I think R is weak.
> 9. I think SG graphics may place R graphics in second place.
>10. I don't think R supports as many graphics formats as SAS.
>11. I don't see as tight an integration with XML that I se with SAS
>R strong points
>The 2000+ packages, which is awsum.
>R does support OOP natively. R has embraced this natively. I am not
>a fan of OOP due to the difficulty I have had interfacing OOP with
>1. Manipulation of matrices, which is non-existent in SAS proper(Base
>SAS). I think this is a huge problem for SAS. I dont see why SAS
>cannot beef up array processing in the datastep. Simple things like X
>[+,+], X[1:10,10], x[*,1] or x[**,1]. Mayde we can do some of this in
>FCMP. I think SAS avoided this to sell more IML licenses?
>2. The flexibilty when dealing with statistical probems. Especiallys
>problems that require linear programming. Newer algorithms that may
>take a year to get into SAS like exact confidence intervals for
>binomial risk differences(9.2). A few lines of R can often give ths
>statistician what he needs. Often a package may pop up in a week or
>3. R makes it easy to produce high quality graphics. I think SAS SG
>graphics will be better. But right now R is looking very good.
>4. Tools to write packages and get them published. This is a big deal
>and is better that Wiki, which seems to a week second for SAS.
>5. I don't see the drive to make money with R like I see with SAS. The
>packages on CRAN are all free and some are really state of the art.
> The R community, like the PERL community is more altruistic and
>6. Until recently R had a much better interface to functions then SAS.
>However the new FCMP, RUN_MACRO, in cojunction with SCL and SAS macro
>languages may level the fieid.
>7. R has better interaces to JAVA
>What I would like SAS to do
>1. Stop spliting up the product. I only have to talk about one R. I
>cannot even keep track of the pieces of SAS.
>2. IML like 'proc sql' should be part of base SAS
>3. Provide a nice R interface in base SAS
>4. Beef up proc report so that rows and columns can easily be merged
>and objects like graphs can occupy the merged areas.
>5. Fix proc report so that when you tell it to order by data it does
>6. Make the JAVA easier to use in Base SAS