Importing sample of large DTA without loading the whole dataset

[Luiza Lins]

I need to import a sample from a .dta file that is too large (67 GB) to load all at once. Let’s say I want to import a random sample of 1% observations from this dataset. Is there any way to do this?

I cannot use:

 use large_file.dta, clear sample 1%, count

Because the program will crash on the first line.

[Rich Goldstein]

-use- has both and "if" and an "in" option so, yes, you can load part of it (including loading only a subset of the variables if you wish); if you need to know the actual size, you can use -describe- to find out using the second syntax in the help file; see

h describe
h use

[daniel klein]

Here is how you use Rich's advice to get a(n approximate) 1 % sample of the observations:

use "large_file.dta" if uniform() < .01