[Thuban-list] visual classifier

[Daniel Calvelo]

I am interested!

BTW I looked into the fortran code you sent me. AFAICT, it's based on
recursive partitionning of the dataset according to a measure of the
goodness-of-fit with straight lines. I'm translating it into Python to test
further; I'll keep the list informed.

I'll try it also with K-means: a long time ago I did that myself. Thanks for
the feedback!!!

Daniel.

-- Daniel Calvelo Aros
-- Dirección General de Información Agraria
-- Ministerio de Agricultura del Perú
-- (51-1) 424-9001


---------- Original Message -----------
From: "Moritz Lennert" <mlennert at club.worldonline.be>
To: "Frank Koormann" <frank.koormann at intevation.de>
Cc: thuban-list at intevation.de, "Daniel Calvelo" <dcalvelo at minag.gob.pe>
Sent: Fri, 5 Dec 2003 10:00:27 +0100 (CET)
Subject: Re: [Thuban-list] visual classifier

> Frank Koormann said:
> > * Moritz Lennert <mlennert at club.worldonline.be> [031204 19:41]:
> >> Daniel Calvelo said:
> >>
> >> >
> >> > I also explored natural breaks. I found very scarce literature on the
> >> > subject.
> >> >>From what I gathered, natural breaks are based on a kernel
> >> approximation
> >> >> to
> >> > the data distribution. From previous experience, programming these
> >> things
> >> > needs tuning and lots of special casing. If you have ideas on the
> >> subject,
> >> > they are most welcome.
> >> >
> >>
> >> We have a fortran program here which was written ages ago and which uses
> >> a
> >> geometrical approach, exploring the slope between data points. If you
> >> are
> >> interested I can send it to you (I will have to check with the authors,
> >> but I would be very surprised if they weren't willing to release the
> >> code
> >> freely).
> >
> > Just for completeness or if some problems arise from using the fortran
> > code: The original paper on natural breaks ("Jenks
> > optimization") is "Error on Choroplethic Maps: Definitions,
> > Measurement, Reduction", G. F. Jenks & F. C. Caspall, Annals of the
> > Association of American Geographers 61 (2), pp. 217-244 (1971)
> >
> > Should be available via JSTOR
> > (http://www.jstor.org/journals/00045608.html)
> > if somebody has an account there.
> >
> 
> I do and I downloaded the article if anyone is interested.
> 
> Moritz
------- End of Original Message -------