[OGRUG] Fwd: [OTT-GOSLING] SAS to R migration

Thu Feb 16 01:03:42 CET 2017

Having once-upon-a-time done some conversions to R from languages other 
than SAS, I would emphasize a few points:

  1/ If this is not strictly automation, that is you have users that do 
ad hoc analysis with the language, then be sure you have a user base 
that is ready to embrace R. Not necessarily everyone, but enough to have 
momentum for the conversion. If they have no programming experience 
other than SAS then you will need to worry a lot about the environment 
you are going to drop them into. License fee savings may be a small part 
of the picture when you start considering training.

2/ If you have a large amount of in-house data then you should probably 
be thinking about R combined with some (SQL?) database. You are almost 
already there if you are using SAS with SQL. But if users have their own 
data in SAS specific rather than SQL format, then you not only have to 
convert the data but you also have to convert the users' way of 
thinking. (The later being much more difficult.)

3/ If you have a lot of automated tasks for which you have scripts that 
need conversion, then my guess is you should be thinking about clean up 
and documentation rather than simply conversion. It is more work in the 
short run, but probably not in the longer run. I would recommend using 
the R package structure for as much of this as you can, because it helps 
you put in reasonable documentation and testing and to think about using 
the scripts in a more supportable/reusable way. Also related, you should 
be using some version control like github, bitbucket, or other. Of 
course, that would be second nature for programmers, but may not be for 
SAS users.

(Sorry I am not really answering your question about organizations that 
have done this from SAS.)

Paul

On 02/15/2017 10:47 AM, Ian E. Gorman wrote:
> The list of URLs given by Glen Newton is a pretty good guide to what is
> needed to convert from SAS to R.
>
> In particular:
> - conversion would be incremental, application by application, and some
> applications might not be worth converting
> - a lot of non-automated work is required
>
> Based on my own experience with large-scale conversion projects that did
> NOT require a major change of implementation language, I think the first
> URL (SAS to R Migration https://www.r-bloggers.com/sas-to-r-migration/)
> gives a very good summary of what a big conversion project would look like.
>
> Conversion between SAS data sets and R data frames should be
> straightforward and it might be possible to fully automate the conversions.
>
> None of the URLs mention the value of tools for working with large
> numbers of text and code files
> - Unix command-line utilities (built into Linux and easily available for
> Mac OS X and Windows 10
> - A third-party file manager (such as ZTreeWin in Windows and Midnight
> Commander in Mac OS X or Linux).
> These tools greatly reduce the cost of non-automated work and greatly
> increase the reproducibility and auditability of the work.
>
>
>
> On 2017-02-15 06:00, r-ug-ottawa-request at r-project.org wrote:
>> Send R-UG-Ottawa mailing list submissions to
>>     r-ug-ottawa at r-project.org
>>
>> To subscribe or unsubscribe via the World Wide Web, visit
>>     https://stat.ethz.ch/mailman/listinfo/r-ug-ottawa
>> or, via email, send a message with subject or body 'help' to
>>     r-ug-ottawa-request at r-project.org
>>
>> You can reach the person managing the list at
>>     r-ug-ottawa-owner at r-project.org
>>
>> When replying, please edit your Subject line so it is more specific
>> than "Re: Contents of R-UG-Ottawa digest..."
>>
>>
>> Today's Topics:
>>
>>    1. Fwd: [OTT-GOSLING] SAS to R migration (Joseph Potvin)
>>
>>
>> ----------------------------------------------------------------------
>>
>> Message: 1
>> Date: Tue, 14 Feb 2017 11:23:57 -0500
>> From: Joseph Potvin <jpotvin at opman.ca>
>> To: "r-ug-ottawa at r-project.org" <r-ug-ottawa at r-project.org>
>> Subject: [OGRUG] Fwd: [OTT-GOSLING] SAS to R migration
>> Message-ID:
>>     <CAKcXiSpFaFr8ertHKt+k8jw=Suazw4Xsf8=0z70W=WatCVqZkA at mail.gmail.com>
>> Content-Type: text/plain; charset="UTF-8"
>>
>> (Is this list still operational?)
>>
>> Forwarding a question from a fellow at a certain science-based federal
>> dept.
>>
>> Joseph Potvin
>>
>> ---------- Forwarded message ----------
>> From: Glen Newton <glen.newton at gmail.com>
>> Date: Tue, Feb 14, 2017 at 11:16 AM
>> Subject: [OTT-GOSLING] SAS to R migration
>> To: GOSLING members in Ottawa <ottawa-gosling at list.goslingcommunity.org>
>>
>>
>> Does anyone on this list know of an example of any GoC or western
>> government organization migrating their users/base from SAS to R?
>>
>> More specifically, when the organization was a science-based organization
>> and the majority of the users being on the science side of things?
>>
>> Just a theoretical question of course...
>>
>> I found some useful resources but no concrete examples as per above:
>> - SAS to R Migration https://www.r-bloggers.com/sas-to-r-migration/
>> - A Saga Of Migrating From SAS To R
>> http://www.mbtmag.com/article/2015/03/saga-migrating-sas-r
>> - Helping Your Organization Migrate to R
>> http://r4stats.com/2015/03/16/migrating-to-r/
>> - SAS to R Migration - LondonR http://www.londonr.org/download/?id=87
>>
>> I am particularly interested where perhaps 90% of the use cases were
>> addressed by R, and perhaps 10% still had to be in SAS, but at least
>> reducing the SAS costs by 90%......
>>
>> Thanks,
>> Glen
>> _______________________________________________
>> Ottawa-gosling mailing list
>> Ottawa-gosling at list.goslingcommunity.org
>> http://list.goslingcommunity.org/mailman/listinfo/ottawa-gosling
>>
>>     [[alternative HTML version deleted]]
>>
>>
>>
>> ------------------------------
>>
>> Subject: Digest Footer
>>
>> _______________________________________________
>> R-UG-Ottawa mailing list
>> R-UG-Ottawa at r-project.org
>> https://stat.ethz.ch/mailman/listinfo/r-ug-ottawa
>>
>> ------------------------------
>>
>> End of R-UG-Ottawa Digest, Vol 41, Issue 1
>> ******************************************
>>
>