101 reputation
bio website
location
age
visits member for 3 years, 2 months
seen Oct 6 '11 at 22:47

Oct
5
comment Switching from C++ to R - limitations/applications
SAS appears faster b/c its data step encourages you to processes data one line at time, whereas R encourages loading everything into memory-- however that isn't the only way to do it. My (biological) datasets are 300gb+ each, and I have no particular problem processing them with a Perl->R pipeline with intermediate data placed in sqlite databases, never loading more than 1gb into memory.