Re: Random Sample - Mailing list pgsql-general

From Reece Hart
Subject Re: Random Sample
Date
Msg-id 1179522973.6910.26.camel@snafu.site
Whole thread Raw
In response to Random Sample  (<tom@tacocat.net>)
List pgsql-general
On Fri, 2007-05-18 at 15:36 -0500, tom@tacocat.net wrote:
> How do I pull a random sample of either 100 records or 5% of the
> population of a table?

If you can be a little flexible about the number of samples, you can try

    select * from table where random()<=0.05;

Of course, there's nothing that guarantees that you'll get 5% and this
only works reasonably for large N. On the other hand, if N were small,
you probably wouldn't be asking for a random sample.

You could also try

    select * from table order by random() limit 100;

That'll be expensive, but get you exactly 100 (if your table has >= 100
rows, of course).

-Reece

--
Reece Hart, http://harts.net/reece/, GPG:0x25EC91A0


pgsql-general by date:

Previous
From: "Martin Gainty"
Date:
Subject: Re: JDBC - Prepared statements and PostgreSql Time/Date operations
Next
From: Tom Lane
Date:
Subject: Re: JDBC - Prepared statements and PostgreSql Time/Date operations