[R] R validation

Marc Schwartz marc_schwartz at comcast.net
Tue Nov 6 22:20:31 CET 2007


On Wed, 2007-11-07 at 09:28 +1300, David Scott wrote:
> In regard to validation of R, there are test directories with test 
> code. See for example:
> https://svn.r-project.org/R/trunk/tests/
> and
> https://svn.r-project.org/R/trunk/src/library/stats/tests/
> 
> David Scott

Those are a part of the tests that are run when using 'make check all'
after building R from source.

As Tony noted, they can be used (and I do) for the IQ (Installation
Qualification) part of the validation process. In my case, I have a
shell script that runs the full process (svn checkout, configure, make
and make check all). The script, using 'mutt', e-mails me the output of
make check all, which has been re-directed to a file, as an attachment,
along with the SVN-REVISION file. I keep that as part of my internal
documentation.

Again as Tony noted, the specifications for the remainder of the OQ/PQ
(Operational Qualification and Performance Qualification) and other
components as required, need to be defined by the organization that
intends to use R in this domain.

As indicated in the R-FDA document, these are areas where the FDA has
provided generally non-prescriptive guidance. As a result, it is up to
each organization to assess its own requirements and implement solutions
that it feels reasonably satisfies those requirements.
 
A starting point might be to review the process currently in place for
similar applications within the organization and assess how they either
do or do not translate for R. You then have the basis for a 'gap'
analysis and can move forward from there.

HTH,

Marc Schwartz



More information about the R-help mailing list