Why distinguish "below detection limit" at the data level? I would use the value measured and enforce the ramification of the detection limit in subsequent analysis. Then use NULL for missing values only.
 ... or isn't that feasible?

Tom
kellert(at)ohsu(dot)edu<mailto:kellert(at)ohsu(dot)edu>
503-494-2442







On Apr 20, 2010, at 12:38 PM, Rich Shepard wrote:

On Tue, 20 Apr 2010, Rich Shepard wrote:

The database we're building uses floating point numbers, where
appropriate. Spreadsheets use whatever someone wants to put in a cell. I
need to look at how NAN is expressed.

  However, this does raise a question that I need to answer: how to
distinguish between a sample taken, analyzed, and with a value below
detection limits and a sample not taken. There may be legitimate reasons for
not taking a sample, or it might have been missed inadvertently. Or, the
sameple might have been taken and analyzed but the results not entered in
the system.

  We need to ponder this.

Rich

--
Sent via pdxpug mailing list (pdxpug(at)postgresql(dot)org<mailto:pdxpug(at)postgresql(dot)org>)
To make changes to your subscription:
http://www.postgresql.org/mailpref/pdxpug

In response to

• Re: Entering '<1' and 'BDL' at 2010-04-20 19:38:31 from Rich Shepard

Responses

• Re: Entering '<1' and 'BDL' at 2010-04-20 20:25:10 from Rich Shepard

pdxpug by date

Next:	From: Rich Shepard	Date: 2010-04-20 20:25:10
	Subject: Re: Entering '<1' and 'BDL'
Previous:	From: Rich Shepard	Date: 2010-04-20 19:38:31
	Subject: Re: Entering '<1' and 'BDL'