[R] Combined variable names

Martin Maechler maechler at stat.math.ethz.ch
Fri Dec 3 09:34:56 CET 2004

>>>>> "Richard" == Richard A O'Keefe <ok at cs.otago.ac.nz>
>>>>>     on Fri, 3 Dec 2004 14:53:04 +1300 (NZDT) writes:

    Richard> I wrote about the perennial "assign to V1 ... Vn" problem:

    >> >What I want to know is *WHY* people are doing this?

    Richard> I failed to make myself clear.

    Richard> What I meant was "what happens NEXT?  Once someone
    Richard> has got past the stage of generating V1 ... V100
    Richard> (or whatever the magic number is), what do they
    Richard> intend to DO with them?

    Richard> I think this is the important question, and I think
    Richard> that this is what the FAQ needs to give help with,
    Richard> and once I understand what people are trying to
    Richard> accomplish *overall* I will be happy to offer some
    Richard> text for the FAQ.

that would be very useful, thank you in advance.

    Richard> I've used stats packages myself that let you
    Richard> abbreviate a range of variables.  The thing is that
    Richard> this was supported by *analysis* methods (and
    Richard> output methods), not just *input* methods.  Now it
    Richard> seems to me that pretty much everything I would
    Richard> want to do with a bunch of separate variables like
    Richard> this in such a package would mean in R that I
    Richard> desperately wanted these things to be columns in a
    Richard> data frame.

or--more generally--named list components. -- important, e.g.,
when the v_j's have differing lengths.

    Richard> Perhaps the next time someone asks this question we
    Richard> can ask them what they intend to do with the
    Richard> variables once they have them.  On past history, we
    Richard> shan't have to wait very long.

good idea (and it may well have been "implemented" occasionaly
in the past).

Martin Maechler, ETH Zurich

More information about the R-help mailing list