[R] reshape speed improvement?

Christian Schulz ozric at web.de
Mon Jun 21 15:03:04 CEST 2004


have anybody a hint how  i can perform my speed for
a big reshape jobs?

My  "long" data.frame have  dim  1.342.116    3    and the "wide" data.frame 
could occur  (if i don't filter the timevar) max. in a 
246744 x  1444 data.frame with a lot of NA what could be set to 0.

Neverthless i didn't need really 1444 columns 
(idependent's for data-mining)  , but  didn't getting
a finish reshape  (for ~100  different values in timevar)  since 
saturday morning on a 1200MHZ Duron-512MB/ R.1.9.1 /Linux.

I check different  min--vsize  and min-nsize  to control the swap, but it 
doesn't help really?

An attempt do this in database with sql doesn't success, too - because 
i recognize that's not trivial construct the columns 
automatic/dynamic like reshape makes it possible.

Thanks for hints/comments

More information about the R-help mailing list