From: | "Sean Davis" <sdavis2(at)mail(dot)nih(dot)gov> |
---|---|
To: | "Mark Roberts" <mailing_lists(at)pandapocket(dot)com> |
Cc: | pgsql <pgsql-sql(at)postgresql(dot)org> |
Subject: | Re: Aggregates with NaN values |
Date: | 2008-12-05 19:08:20 |
Message-ID: | 264855a00812051108sc042f99t62703ba1fa07d2f5@mail.gmail.com |
Views: | Raw Message | Whole Thread | Download mbox | Resend email |
Thread: | |
Lists: | pgsql-sql |
On Fri, Dec 5, 2008 at 1:51 PM, Mark Roberts
<mailing_lists(at)pandapocket(dot)com> wrote:
>
> On Thu, 2008-12-04 at 13:01 -0500, Sean Davis wrote:
>> I am happy to see NaN and infinity handled in input. I would now like
>> to compute aggregates (avg, min, max, etc) on columns with NaN values
>> in them. The standard behavior (it appears) is to have the aggregate
>> return NaN if the data contain one-or-more NaN values. I am used to
>> using coalesce with NULL values, but that doesn't work with NaN. I
>> can deal with these using CASE statuement to assign a value, but is
>> there a standard way of dealing with the NaN (or Infinity, for that
>> matter) cases to get a behvavior where they are "ignored" by an
>> aggregate?
>>
>> Thanks,
>> Sean
>>
>
> Have you considered using a where clause?
Thanks, Mark. Yes. I have about 20 columns over which I want to
simultaneously compute aggregates. Each has NaN's in different rows,
so a where clause won't do what I need.
The CASE statement approach works fine, though.
Sean
From | Date | Subject | |
---|---|---|---|
Next Message | Oliveiros Cristina | 2008-12-05 19:23:25 | Re: Best way to "and" from a one-to-many joined table? |
Previous Message | Bryce Nesbitt | 2008-12-05 18:55:23 | Best way to "and" from a one-to-many joined table? |