[GRASS5] Re: d.vect.thematic wish: handle null values

[Michael Barton]

Moritz,

This is great.

Michael
______________________________
Michael Barton, Professor of Anthropology
School of Human Evolution and Social Change
Arizona State University
Tempe, AZ  85287-2402
USA

voice: 480-965-6262; fax: 480-965-7671
www: http://www.public.asu.edu/~cmbarton


> From: Moritz Lennert <mlennert at club.worldonline.be>
> Date: Tue, 30 Aug 2005 15:25:00 +0200
> To: Daniel Calvelo <daniel.calvelo at minag.gob.pe>
> Cc: <grass5 at grass.itc.it>, <michael.barton at asu.edu>
> Subject: Re: [GRASS5] Re: d.vect.thematic wish: handle null values
> 
> -----BEGIN PGP SIGNED MESSAGE-----
> Hash: SHA1
> 
> Daniel Calvelo Aros wrote:
>> From: "Moritz Lennert" <mlennert at club.worldonline.be>
>> Sent: Sun, 28 Aug 2005 11:53:34 +0200 (CEST)
>> 
>>>> When I was dealing with this problem I only used the DBF driver,
>>>> which gave zeros for nulls under unspecified cirsumstances.
>>> 
>>> Thinking about this a bit more this seems more like a bug to me.
>>> Making nulls into zeroes gives very different results than if nulls
>>> are excluded from the calculations.
>>> 
>>> At the user should be warned about this.
>>> As I mentioned earlier, we have to think a about whether it is
>>> better to silently exclude nulls, or whether we should explicitely
>>> give the opportunity to exclude a certain value (or range or values)
>>> from the classification. This would then allow those with nulls in
>>> their files to transform them to any value (e.g. -99999) and exclude
>>> it from the calculations.
>>> 
>>> 
>>>> I can fix the script to filter out nulls, but I'm pretty sure the
>>>> problem will
>>>> remain under certain configurations.
>>> 
>>> Maybe above solution is more generic. But in any case nulls should
>>> be handled more gracefully...or at least a warning included in the
>>> man page...
>> 
>> 
>> I fixed CVS to *ignore* nulls.
> 
> Great, now it works for me. Thanks !
> 
>> 
>> I'd say that is a good choice for use in d.vect.thematic. If you really
>> really
>> want to use a symbol/color for NULL values, there is always the possibility
>> of
>> overlaying d.vect.thematic with a d.vect where="the_field is null".
>> 
>> Furthermore, current db.execute allows for things like 'update table set
>> field=cat+23 where field is null', which allows to take care of any case I
>> can
>> think of, either inserting new fields and marking them appropriately or
>> modifying existing fields in-place.
> 
> I agree.
> 
> BTW: I just committed a slight addition to the legend drawing
> routine for drawing black lines around the legend boxes. Just
> thought it looks nicer that way...now if only these legends could be
> exported in a more usable form...(will commit a wish for this just
> for memory)
> 
> Moritz
> -----BEGIN PGP SIGNATURE-----
> Version: GnuPG v1.4.1 (GNU/Linux)
> Comment: Using GnuPG with Thunderbird - http://enigmail.mozdev.org
> 
> iD8DBQFDFF4srIrMbm76jD8RAstWAJ9MM6l3oLGxBfLYQNxtwUMw5Pj4agCeKPbO
> lOXYOACxZdQwlciec5Z2rsY=
> =nS1Z
> -----END PGP SIGNATURE-----