[R] Newbie questions

Zsombor Cseres-Gergely z.cseres-gergely at ucl.ac.uk
Tue Oct 31 20:49:31 CET 2000

On Mon, Oct 30, 2000 at 03:07:48PM -0800, Thomas Lumley wrote:

> Yes it is unavoidable though it isn't because of object
> orientedness. Stata saves memory by allowing only one rectangular dataset
> at a time, which simplifies things a lot. Stata also has a simpler
> programming language. 

But should this affect memory usage when using more than data frame? As far I
have experienced, the memory need (for loading the data) of Stata is about the
size of the file on disk, but R consumes more that twice that. This shuld not
be connected to the language or the ability to accept more datasets, but the
representation of the data, should it?

> R is scalable up to at least 2Gb of memory (perhaps more now on 64bit
> machines).   It does not scale with added CPUs. Faster CPUs help, of
> course.

Linearly (both?)?

> library(ctest) is in the main R distribution. It has a lot of classical
> hypothesis tests.

Well, I was not that precise, since I meant tests for regression parameters.
Now I have seen from the code of ellipse(), that one can extract the varcovar
matrix of the extimators as summarl.lm$cov.unscaled, and than the calculations
are straightforward (at least for simple tests). But I still wonder if there
is no library for this, since (ctest) does not do this, AFAIK.

> It's good to hear that R works with the Hurd.

Yeah, pure GNU system builds on my machine. Since nobody answered the "how
about survey" question, I assume that it will be implemented when I do it.
After this is done, I can replace Stata, I think.

Thank you,

PS: In fact: thank you for all. I think this e-mail was the longest. Now I am
on the list, so no need for cc:!
r-help mailing list -- Read http://www.ci.tuwien.ac.at/~hornik/R/R-FAQ.html
Send "info", "help", or "[un]subscribe"
(in the "body", not the subject !)  To: r-help-request at stat.math.ethz.ch

More information about the R-help mailing list