"Marc G. Fournier" <scrappy(at)hub(dot)org> writes:
> n_distinct, I'm taking it, of -1 means its a UNIQUE INDEX, and the
> positive numbers are exact #'s ... but, -0.976321? :)
"Almost unique", evidently. See stadistinct at
http://www.ca.postgresql.org/users-lounge/docs/7.2/postgres/catalog-pg-statistic.html
> Also, how is the avg_width interpreted?
Just what you'd expect: average stored width, in bytes, of non-NULL
entries in the column.
regards, tom lane