Date: Wed, 13 Jul 2011 08:58:10 -0500
From: "ktm@rice.edu" <ktm@rice.edu>
To: "Jira, Marcel" <Marcel.Jira@wu.ac.at>
Cc: "'pgsql-sql@postgresql.org'" <pgsql-sql@postgresql.org>
Subject: Re: Select random lines of a table using a probability
 distribution
Message-ID: <20110713135810.GA1874@staff-mud-56-27.rice.edu>
References: 
 <D793F5C522F1DD40BB9DC43586C57637E098CC87C6@MBX-B.ad.wu-wien.ac.at>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: 
 <D793F5C522F1DD40BB9DC43586C57637E098CC87C6@MBX-B.ad.wu-wien.ac.at>
User-Agent: Mutt/1.5.20 (2009-12-10)

On Wed, Jul 13, 2011 at 03:27:10PM +0200, Jira, Marcel wrote:
> Hi!
> 
> Let's consider I have a table like this
> 
> id    qualification    gender    age    income
> 
> I'd like to select (for example 100) lines of this table by random, but the random mechanism has to follow a certain probability distribution.
> 
> I want to use this procedure to construct a test group for another selection.
> 
> Example:
> 
> I filter all lines having the qualification "plumber".
> I get 50 different ids consisting of 40 males, 10 females and a certain age distribution.
> 
> I also get some information concerning the income of the plumbers.
> 
> Now I want to know if the income is more influenced by the gender and age distribution or by the qualification "plumber".
> 
> Therefore I would like to select a test group (of 50 or more) without any plumbers. This test group has to follow the same age and gender distribution.
> 
> Then I would be able to compare this groups income statistics with the plumbers income statistics.
> 
> Is this possible (and doable with reasonable effort) in PostgreSQL?
> 
> Thank you in advance.
> 
> Best regards,
> 
> Marcel Jira
> 

You may want to take a look at pl/R which make the R system available to
PostgreSQL as a function language.

Regards,
Ken