On 2012-07-02 12:25, David Winsemius wrote:
On Jul 2, 2012, at 2:21 AM, Bert Gunter wrote:
Hi Kristina:
That's because I gave you the wrong function. It should be "with" not
"within", as:
sort2v4$adj_mean<-with(sort2v4, ave(mean, point, FUN=function(x)x-
x[1]))
within(dat,...) returns dat suitably modified.
with (dat,...) just returns the result of the function.
I frequently use within() when I mean with(), unfortunately. Sorry
for the
error.
That's interesting to me. I almost never use `within`. I'm wondering
if I'm missing something? As I read the help page `within` is going to
do something different when the expression is an assignment, so one
could do the following with `within`
within(dfrm, var2 <- var2/100)
... but that would have no lasting effect since its result was not
assigned, so I do not se an advantage over:
dfrm$var2 <- with(dfrm, var2/100)
In both case one would still need to assign the result to an object,
in the case of the first to 'dfrm' itself and in the second case to
dfrm[['var2']].
None of the examples on the help page use `within`. Is that because it
is inferior to transform?
[....]
I don't think that within() is inferior to transform().
I started using it after seeing some code by Bill Venables
and I now use it fairly frequently (not as frequently as the
quite useful with(), though). I think that within() can be
more versatile than transform(); consider:
aq <- head(airquality)
aq1 <- transform(aq, new1 = Ozone^2,
new2 = new1 + 10)
#Error in eval(expr, envir, enclos) : object 'new1' not found
aq1 <- within(aq, {new1 <- Ozone^2;
new2 <- new1 + 10})
I agree that, for a single new or redefined variable,
the 'df$newvar <- ...' assignment is straightforward,
but for fairly extensive modifications, within() is
very nice. The only thing that bugs me about within()
is that the new variables are appended in "reverse"
order.
Peter Ehlers
______________________________________________
R-help@r-project.org mailing list
https://stat.ethz.ch/mailman/listinfo/r-help
PLEASE do read the posting guide http://www.R-project.org/posting-guide.html
and provide commented, minimal, self-contained, reproducible code.
