Re: median for postgresql 8.3

Поиск
Список
Период
Сортировка
От Brent Wood
Тема Re: median for postgresql 8.3
Дата
Msg-id 4CE3A6210200007B00032B51@gwia.niwa.co.nz
обсуждение исходный текст
Ответ на median for postgresql 8.3  (maarten <maarten.foque@edchq.com>)
Ответы Re: median for postgresql 8.3  (maarten <maarten.foque@edchq.com>)
Список pgsql-general
Hi Maarten,

The best way I know of to do this is not to do statistical queries "in" the DB at all, but use a stats capability embedded in your database, so they still appear to the user to be done in the db. I don't see how you can easily get the functionality you want without user defined functions or addons, While PL/R is a "special addon", and you created a custom median function to do this, there are very good instructions to follow to do this.

 I think it might be worth your while if you are looking to retrieve stats from SQL queries.

See PL/R, and the median how-to at:
http://www.bostongis.com/PrinterFriendly.aspx?content_name=postgresql_plr_tut01

HTH,

  Brent Wood



Brent Wood
DBA/GIS consultant
NIWA, Wellington
New Zealand

>>> maarten <maarten.foque@edchq.com> 11/17/10 9:15 AM >>>
Hello everyone,

I was doing some analysis of data to find average delays between some
timestamp values etc...
When the number of rows the average is computed over is small, this can
give distorted values. So I've obviously added a count column to see if
the average represents much data.
However, I would also like to add the median value to give me a pretty
good idea of whats happening even for smaller counts.

I couldn't find such an aggregate function in the manual (version 8.3)
and some websearching didn't uncover it either.

I was thinking about
SELECT max(id) FROM test ORDER BY id ASC LIMIT
    (SELECT count(*)/2 FROM test)

But two things are wrong with that:
Limit can't use subqueries :(
And ORDER BY gives me the error: 'must be used in aggregate function
etc...) but I can probably work around this by using an ordered subquery
in stead of the table directly.

Furthermore, I need the median for a timestamp column, which would
probably complicate things more than when it is a number column.

I'd like to be able to do this using only the database. (So no
programming functions, special addons etc...)

Any ideas anyone?

regards,
Maarten


--
Sent via pgsql-general mailing list (pgsql-general@postgresql.org)
To make changes to your subscription:
http://www.postgresql.org/mailpref/pgsql-general

Please consider the environment before printing this email.

NIWA is the trading name of the National Institute of Water & Atmospheric Research Ltd.

В списке pgsql-general по дате отправления:

Предыдущее
От: Peter Geoghegan
Дата:
Сообщение: Re: How to select a postgresql table inside Oracle
Следующее
От: Vick Khera
Дата:
Сообщение: Re: AfterTriggerSaveEvent() called outside of query