Archives of the TeradataForum
Message Posted: Fri, 26 Mar 2004 @ 21:02:37 GMT
| Subj: || || Re: SAMPLE syntax |
| From: || || Victor Sokovin |
| ||I don't see anything uncertain if you just recognize that the wording in the documentation is wrong.|| |
If you recognize that, then nothing is uncertain. Otherwise, you might think you are doing something wrong or you see a bug. I used the word
"uncertainty" to describe that situation.
| ||Proof: You can't get 0% of the rows nor can you get 100% (those are the 2 endpoints).|| |
I think a few would care about 0% samples (except perhaps code generators) but 100% samples are not that useless. Teradata offers random
samples of good quality so 100% sample might mean: "give me the whole table in random order". Michael's suggestion to use "SAMPLE 0.5, 0.5" would
probably work for the OP but not for somebody who would be interested in random order on the whole table and cares about mathematical statistics.
Well, I think this goes beyond the current thread, so I agree with you that enough have been said.