[R-users] Some R stuff from the JSM Meeting in San Fransisco

[Matt Pocernich]

Hi,

I got to go the Joint Statistical Meeting in San Fransisco last week.  The
following are a couple notes relating to R.

M

*****
The session titled the State of Statistical Software, was basically a
cult meeting of R Software users.  The first speaker Jan D Leeuw -
editor of the Journal of Statistical Software (www.jstatsoft.org) .  He
began the talk
making some of the following statements.  (nearly verbatim).
"R is the lingua franca of statistical research.  Work in all other
languages should be discouraged."  "SAS
is something that statisticians should know about, but not actually
use."

Duncan Temple (from Bell Labs and the Omegahat Project) more or less
said, that much of the code developed today will be lost in the next
10 years.  In part
this is due to poor programming quality and the failure of statisticians
to get outsiders to use it.  To address the first concern, he pressed
the need for an equivalence of journal reviewed software.
Additionally, academic people should be given more credit, career wise,
for developing software since in many cases, its more useful than
journal articles.  On the second issue, he first stated that people in
other fields are not going to willingly learn another language.  To
try and make them do so would be difficult, maybe impossible.  To
facilitate interactions with outside fields, he presented ways in which
S (and R) can be connected with other software.  With some effort,
this can be done so that non-R users should not have to learn R.

To support the practice of better software development, Luke Tierney
presented  three new R features.  These included name space
management, advanced error handling and dynamic variables. These
features are included in the methods library which is part of the
basic installation, in version 1.7.  (I think
these are discussed in the most recent issue of R-Notes.)

There was also a software engineer from SAS.  He was politely
received.

On the topic of R software, but not a part of this session, Paul
Murrell presented a early version of a package called gridSVG.  SVG is
a file format viewable by Adobe Acrobat.  By by-passing R's basic
device definition, it allows some interesting things to be done,
particularly in a web environment.  Links can be embedded within a
graph.  For example, a link embedded in an axis may take the user to
another graph.  Additionally, one can zoom in on specific areas of the
graph.  Animation is also possible with this package.  He presented
some simple examples such as a series of ball bouncing in the gravity
for different planets.


Matt Pocernich
NCAR - Research Applications Program
303-497-8312