1

I loaded big data files (https://www.kaggle.com/c/avazu-ctr-prediction) using the read.csv.ff command from ff package using the following command:

 train = read.csv.ffdf(file="path to my big data files/train.csv",VERBOSE=TRUE)

then save it using the usual save workspace from RStudio. It created two files:

an .RData file and .ffData one which is a compressed file.

Everything worked fine but each time I restart my computer or close RStudio and open RStudio again, I'm unable to access ffdf objects from my workspace giving the error:

 >train[1,]

    file.access(filename, 0) == 0 is not TRUE

Or I can see the ffobject in the workspace I loaded.

What is this problem? how could I fix it?

Thanks in advance, HR

Herimanitra Ranaivoson
  • 51
  • 2
  • 7
  • Please read the posting guidelines. You have not provided your code, so we have no idea what `train` is or how you built it. I'm making an unjustified guess that your `train` is an `ff` object, in which case rather obviously you either deleted the `.ffData` file it belongs to, changed its permissions, or did some other bad thing to it. – Carl Witthoft Nov 29 '14 at 13:09

1 Answers1

0

Use 'ffsave', not 'save', and then 'ffload' when you want to read it back in again. So, something like:

ffsave(train,file = 'mynamedtrainfile)
HywelMJ
  • 332
  • 2
  • 7