[R] Manage huge database

Barry Rowlingson b.rowlingson at lancaster.ac.uk
Mon Sep 22 17:46:52 CEST 2008


2008/9/22 jim holtman <jholtman at gmail.com>:
> Why don't you make one pass through your data and encode you
> characters as integers (it would appear that you only have 16
> combinations).  You might also want to consider using the 'raw' object
> since these only take up one byte of storage -- will reduce your
> storage requirements by 4.  Then store each row in a 'filehash' object
> so you can quickly retrieve a row at a time and then index directly to
> the byte(s) that have the information that you want.

 My original response of specifying a relational database now seems
somewhat comical :)

Barry



More information about the R-help mailing list