3

I have a dataframe that has 4755 rows and 27199 columns. It's actually a document term matrix and I'm trying to perform feature selection using the "FSelector" package. Here is some of the code below:

library(FSelector)
weights <- information.gain(Flag~., dtmmatdf)

Each time I do this I get an error

Error: protect(): protection stack overflow

I have a 24GB RAM and the dataframe is about 500Mb in size. So I don't know what the problem is and how do I fix it?

Community
  • 1
  • 1
anonymous
  • 405
  • 8
  • 22
  • Try `R --max-pp-size=100000`. Also it might be a bug in `FSelector` – romants Aug 24 '15 at 16:49
  • I think there seems to be a bug in FSelector, but how do I fix it? – anonymous Aug 24 '15 at 19:03
  • Write to developers of `FSelector`? https://cran.r-project.org/web/packages/FSelector/index.html The project doesn't seem to be open-source, which makes it significantly harder to fix. You can try debugging line by line to see where an error is, maybe you will get lucky – romants Aug 24 '15 at 22:54
  • The repo is public at https://github.com/larskotthoff/fselector -- can you post a complete example that reproduces the problem please? – Lars Kotthoff Oct 06 '15 at 00:12

0 Answers0