|
|
Archives of the TeradataForum
Message Posted: Fri, 03 Dec 2004 @ 15:51:55 GMT
Subj: | | Re: Best way of writing SQL to retrieve a sample |
|
From: | | Michael Larkins |
Although a function like RANK will work for this, it is not a random sample. Instead, it will always return the highest values or optional the
lowest values. Plus, you did not tell us which release of Teradata you are running, so here a way to get a random sampling in both V2R4 and
V2R5:
sel country_code, state_code, county_code
from (sel country_code, state_code, county_code, csum(1,1) as randomizer
from country_table group by 1 qualify randomizer < 51) DerivedTable
If you have V2R5 and prefer ANSI standard SQL, you should be able to do the same thing with ROW_NUMBER.
Hope this helps,
Michael Larkins
Certified Teradata Master
Certified Teradata SQL Instructor
| |