[Thuban-list] Patch: classifiers

[Daniel Calvelo Aros]

On Tue, 10 Feb 2004 16:03:20 +0100 (CET), Moritz Lennert wrote
[...]
> I just (superficially) tested the patches and the classifiers are 
> just great. Just one remark concerning the discontinuities 
> classifier: would it be possible to let the user chose the number of 
> classes she wants ? The original code from J.P. Grimmeau gives the 
> solutions for a number of different classes. Ideally the use should 
> be given a choice of the number of classes based on a list of the 
> potential breaks. At the current state, I can obviously fusion two 
> classes, but I do not have the info on which break is the smallest...

Ok, I'll work on that. But the original code DEFINES the number of breaks.
It's part of its strength. Giving you an indicator of the overall quality for
the visual classifier would be helpful, I reckon. Let me think around it a bit.

> Concerning the hsv patch, I don't seem to understand how it works. At
> least when I chose the extreme colors and then click on "Generate" I 
> get exactly the same colors, whatever the chosen mode.
> 
> Moritz

Both methods give quite similar results for e.g. black to blue, white to red,
but very different for, e.g. red to green. Try that one. In principle, ramps
that vary mainly in saturation or value should be similar. If hue changes are
involved, the HSV classifier moves along the spectrum, in the "closest
direction" (e.g. red to blue goes through purple but red to green goes through
yellow, not purple and blue).

Daniel.

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