Hi Pierre, How do you know in which group each row belongs? If you don't care how the rows are grouped, you can say this: create table foo (v float); insert into foo select random() from generate_series(1, 1000000) s(a); select n % 50 g, stddev(v) from (select row_number() over () n, v from foo) x group by g; On the other hand if you have some way of ordering the rows you could say this: create table foo (id integer, v float); insert into foo select a, random() from generate_series(1, 1000000) s(a); select (n - 1) / 50 g, stddev(v), count(*) from (select row_number() over (order by id) n, v from foo) x group by g order by g; Yours, Paul On Thu, Jan 22, 2015 at 7:18 AM, Pierre Hsieh <pierre.hsieh@xxxxxxxxx> wrote: > Hi > > This table just has a column which type is integer. There are one million > data in this table. I wanna calculate standard deviation on each 50 data by > order. It means SD1 is from data 1 to data 50, SD2 is from data 51 to > 100.... Is there anyone who can give me some suggestions? Thanks > > Pierre -- _________________________________ Pulchritudo splendor veritatis. -- Sent via pgsql-general mailing list (pgsql-general@xxxxxxxxxxxxxx) To make changes to your subscription: http://www.postgresql.org/mailpref/pgsql-general
