 |
 |
Archives of the TeradataForum
Message Posted: Thu, 06 Nov 2003 @ 14:16:44 GMT
| Subj: | | Re: Using the Sample Clause? |
| |
| From: | | Geoffrey Rommel |
| | > Has anyone checked the statistical validity of SAMPLE clause? | |
Some years ago, when we were on V2R3, our statisticians asked this very question. They were eventually satisfied that the sampling
was acceptable, but I don't know what criteria they used. The bottom line is that SAMPLE and RANDOM apparently use lrand48, which is
generally considered pretty good; the algorithm is given on the man page for that function. If you want to know, however, whether the
samples are appropriate for your application, you will have to run your own statistical tests (chi squared, spectral, etc.).
| |
