Splitting columns when using ddply

Question

I'm using ddply and am stuck with the way the output is arranged. This is the code I'm using. As you can see, the final output (timeseries.out) has the original data and predicted data in one column.

data <- data.frame(Product = c(rep("Shampoo",5),rep("Soap",5)),
               TSdata = rnorm(10, 1, 10))

tsfun <-function(y){
  arima.out <- arima(y$TSdata)
  arima.fc <- predict(arima.out, n.ahead=5)
  return (data.frame(c(y$TSdata, arima.fc$pred)))
}

library(plyr)
timeseries.out <- ddply(data, .(Product), tsfun)

What I really want is the original data in one column, and the predicted data in another column with NAs filling in the blank spots.

data.out <-data.frame(Product = timeseries.out[1:10,1],
       Data = c(timeseries.out[1:5,2], rep("NA",5)),
       Forecast = c(rep("NA",5),timeseries.out[6:10,2]))

How can I change the returned value from tsfun so it looks like data.out? I've tried a number of things, but either get errors or the wrong result.

Thanks!

Answer 1

By changing the return value I was able to get the output I wanted. Thanks to joran for getting me thinking on the right path.

Simple modification of the return statement so two columns are output instead of one

return (data.frame(y$TSdata, as.numeric(arima.fc$pred)))

Return value modified to fit desired output of two columns with NAs

return (data.frame(c(y$TSdata, rep(NA, length(arima.fc$pred))),
                   c(rep(NA, length(y$TSdata)), arima.fc$pred)))