[R] ERROR : cannot allocate vector of size (in MB & GB)
Sarah Goslee
sarah.goslee at gmail.com
Tue Jul 24 19:59:06 CEST 2012
Sure, get more RAM. 2GB is a tiny amount if you need to load files of
1GB into R, and as you've discovered won't work.
You can try a few simpler things, like making sure there's nothing
loaded into R except what you absolutely need.
It looks like there's no reason to read the entire file into R at once
for what you want to do, so you could also load a chunk, process that,
then move onto the next one.
Sarah
On Tue, Jul 24, 2012 at 9:45 AM, Rantony <antony.akkara at ge.com> wrote:
> Hi,
>
> Here in R, I need to load a huge file(.csv) , its size is 200MB. [may come
> more than 1GB sometimes].
> When i tried to load into a variable it taking too much of time and after
> that when i do cbind by groups,
> getting an error like this
>
> " Error: cannot allocate vector of size 82.4 Mb "
>
> My requirement is, spilt data from Huge-size-file(.csv) to no. of small csv
> files.
> Here i will give no of lines to be 'split by' as input.
>
> Below i give my code
> -------------------------------
> SplitLargeCSVToMany <- function(DataMatrix,Destination,NoOfLineToGroup)
> {
> test <- data.frame(read.csv(DataMatrix))
>
> # create groups No.of rows
> group <- rep(1:NROW(test), each=NoOfLineToGroup)
> new.test <- cbind(test, group=group)
> new.test2 <- new.test
> new.test2[,ncol(new.test2)] <- NULL
>
> # now get indices to write out
> indices <- split(seq(nrow(test)), new.test[, 'group'])
>
> # now write out the files
> for (i in names(indices))
> {
> write.csv(new.test2[indices[[i]],], file=paste(Destination,"data.", i,
> ".csv", sep=""),row.names=FALSE)
> }
> }
>
> -----------------------------------------------------
> My system Configuration is,
> Intel Core2 Duo
> speed : 3GHz
> 2 GB RAM
> OS: Windows-XP [ServicePack-3]
> ---------------------------------------------------
>
> Any hope to solve this issue ?
>
> Thanks in advance,
> Antony.
>
>
>
>
> --
--
Sarah Goslee
http://www.functionaldiversity.org
More information about the R-help
mailing list