思路:avg()over
select
id
,job
,score
from
(
select
*
,avg(score)over(partition by job) av_g
from grade
) a
where score > av_g
order by 1