Gareth Campbell wrote:> Hello,
> I have a matrix and I want to sample 20 rows that are the the percentiles
of
> 0-100 in 0.05 increments. I have a vector of my sequence
>
> (0, 0.05, 0.10, 0.15,....1.0) and also
>
That's actually 21 entries, but that probabl doesn't
matter...> a normalised vector of rownumbers. That is, there are 234 rows (for
> example) so I do
>
> perc<-c(1:234/234)
>
> which looks like a bunch of numbers from 0 - 1.
>
> In Excel (which I try not to use at every possible occaison) you can use a
> VLOOKUP function to say choose the rows from the perc vector that are the
> closest match to the sequence vector - i.e. you can specify to choose the
> "closest" match, it doesn't have to be an exact match.
>
> In R, I'd normally use the grep command when I know it's an exact
match.
> Does anyone know how to achieve this in R?
>
You could do it with approx. That is:
rownum <- approxfun( (1:234 - 0.5)/234, 1:234, method="constant",
rule=2)
Then rownum(0.05) will give 12, rownum(0.1) will give 23, etc.
Duncan Murdoch> Also, I've been getting heaps of help from everyone on this mailing
list -
> if I reply to the respondent and R Help again, will it show up in the
> correct thread? I haven't been able to say thankyou as I wasn't
sure it
> would show up in the thread.
>
> Thanks
>
>
>