Random samples from R data frames.

There are times you just have too much data, random samples are nice to test assumptions and algorithms first.

So in R you can create a function to return a random sample of a data frame for such emergencies.

randomSample = function(df,n) { 
   return (df[sample(nrow(df), n),])
}

And to use:

smallerDF<-randomSample(bigDF, 40)

(40 being the number of rows you want in your sample).

Jason Bell

Phillip Burger says:

February 2, 2014 at 2:23 am

Thanks for the post! I was able to use this function. Strange that I hadn’t encountered previously the need to randomly sample observations from a data frame?!

To other readers, make sure you use the appropriate value for your replace = argument.I think for sampling a data frame, most often the default of replace = FALSE is probably going to be what you want.
Jason Bell says:

February 2, 2014 at 9:11 am

Hi Philip,

Thanks for taking the time to comment. I don’t work with small data sets so the need to create a small random sample to test an algorithm is probably more than most. And you’re right about the FALSE statement, nice catch!

Regards
Jase
HeinrichHeinrich says:

April 21, 2014 at 2:59 pm

Note that the above method does not preserve the ordering of the rows. Use sort(sampe(..)) if this is desired.