[Rd] Bridging R to OpenOffice

Leonard Mada lmada at gmx.net
Wed Mar 28 20:29:18 CEST 2007


Many thanks for the many kind replies. It is very reassuring to have 
support from a strong community.


Hin-Tak Leung wrote:
 > Hmm, if all you are interested is reading/writing Excel spreadsheets
 > from R, there are much lighter and easier ways of doing it, than
 > hooking up with openoffice. The Perl people have had
 > Spreadsheet::ParseExcel and Spreadsheet::WriteExcel for years (and
 > they work quite well, personal experience). Those are tiny
 > (a couple of Mb's?) compared to the size of openoffice.

I believe that this R-OOo bridge should pursue a different path. I 
favour the idea to facilitate access to R for common spreadsheet users. 
As these users are less likely to learn the full S language, the 
implemented method should by largely offer a GUI-driven interface to 
important statistical (R)-functions (at least in the beginning; adding 
further functionality later on).

Having an R package to read/write .ods files seems reasonable, too, (and 
I would definitely like it) however this will not benefit the larger 
spreadsheet community. Again, it will ease the life of power users, but 
the novice must still first learn R. The package odfWeave (see R News 
vol 6/4, October 2006) offers already basic support for OOo Writer files 
and, while it currently lacks spreadsheet functionality, I am looking 
forward to see it implemented, too.

1. Teaching Role
There are some deeper reasons why I cling to the R-OOo bridging idea. I 
have read in my life hundreds of biomedical articles (probably even more 
than a thousand) and I have a very bitter taste about the quality of 
most of these articles. The statistics have played an important role in 
my judgment.

The fact is, that most researchers will use a spreadsheet program to 
gather their data. And most will use this spreadsheet program to do 
their analysis, too. If this spreadsheet program offers more advanced 
statistical methods (and also a sensible help file on these methods), 
then some users will try to use them. Some of these will take the next 
step, too, and will dwell a little bit deeper into statistics, thus 
raising the quality of the research.

In this way, this bridge would have also a teaching role, persuading 
some users to take a deeper look at statistics (especially learning more 
advanced and various newer methods). It will make R more popular, too.

2. Implementing Advanced Statistical Functions in OOo
I do not favour this idea, because:
- newer methods are not always trivial to implement
- spreadsheet programs are notorious for poor statistical algorithms 
(non-robust implementation)
- more resources (programmers, testing frameworks) are needed, when free 
(and much better) alternatives already do exist
- community would have to form first (e.g. help, FAQ), while R already 
has a large community

Many thanks,

Leonard



More information about the R-devel mailing list