Is it possible to use write.table() and ddply, together?

Question

Let's say I have a data.frame like:

a <- c(1:10,1:10,1:10,1:10,1:10,1:10,1:10,1:10,1:10,1:10)
df <- data.frame(a,rnorm(100))

And I want to be able to write a csv file for each value of x. Is it possible to do this with ddply?

I can already do this with a for loop in a few lines... but I'm curious if it's possible to do it with ddply.

for (x in 1:nrow(unique(df["a"]))) {
tmp <- unique(df["a"])
tmp2 <- paste(tmp[x,],".csv", sep="")
write.table(subset(df, a == tmp[a,], drop=T),file=tmp2, sep=",", row.names=F)
}  

Answer 1

It's possible to do it with ddply, but that is not what the function was designed for. From the plyr documentation:

All plyr functions use the same split-apply-combine strategy...

You want to split the data.frame and apply a function but you don't want to return anything, so ddply will throw an error if you don't return something that can be combined into a data.frame.

Answer 2

Continuing from Joshua's answer, the plyr function to use is d_ply which does not expect to return anything. You can do something like this:

d_ply(df, .(a),
      function(sdf) write.csv(sdf,
                              file=paste(sdf$a[[1]],".csv",sep="")))

The file argument to write.csv is constructed such that each subset gets a different filename.