Ron Mayer <rm_pg(at)cheapcomplexdevices(dot)com> writes:
> What I think I'd like to see is for there to be
> another statistic similar to "correlation" but rather
> than looking at the total-ordering of the table, to
> look how correlated values within any single page are.
Our current approach to correlation is surely far too simplistic; it was
just a quick hack to get *something* in there for this consideration.
I don't personally know enough about statistics to do much better :-(
regards, tom lane