[mdlug] simple stats from piping numbers into R?

[Robert Citek]

On Mon, Apr 19, 2010 at 8:57 PM, Michael Corral <micorral at comcast.net> wrote:
> You could do it like this:
>
> R --vanilla --slave --args $(seq 1 100) < stats.r
>
> where the file stats.r consists of these two lines:
>
> num <- as.numeric(commandArgs(TRUE))
> summary(num)

Thanks, Michael.  That'll work in a pinch for small datasets.  Looks
like for larger sets I'll have to find some other way.

One way I've been exploring is piping a script to R, using bash's
here-document feature with variable interpolation.  Works even for 1
MM numbers:

$ cmd="seq 1 1000000"  ; { cat <<eof
summary(read.csv(pipe("$cmd"), head=FALSE))
eof
} | time -p R --slave
       V1
 Min.   :      1
 1st Qu.: 250001
 Median : 500000
 Mean   : 500000
 3rd Qu.: 750000
 Max.   :1000000
real 12.50
user 15.98
sys 0.26

I can then substitute various commands for the cmd variable.  Not
ideal, but will work.

Regards,
- Robert