Re: ANALYZE sampling is too good

On Tue, Dec 10, 2013 at 4:48 PM, Peter Geoghegan <pg(at)heroku(dot)com> wrote:
> Why would I even mention that to a statistician? We want guidance. But
> yes, I bet I could give a statistician an explanation of statistics
> target that they'd understand without too much trouble.

Actually, I think that if we told a statistician about the statistics
target, his or her response would be: why would you presume to know
ahead of time what statistics target is going to be effective? I
suspect that the basic problem is that it isn't adaptive. I think that
if we could somehow characterize the quality of our sample as we took
it, and then cease sampling when we reached a certain degree of
confidence in its quality, that would be helpful. It might not even
matter that the sample was clustered from various blocks.

--
Peter Geoghegan