|
|
"Kenneth" <kdw### [at] gmailcom> wrote:
> I had to look up that term (I already knew the concept, but not the name; thanks
> for the tip.)
I remember them from an HPLC course I took circa 1993.
They used them to plot mobile phase composition.
> To do adequate justice to media-- its many possible variables--a ternary plot
> would not be adequate; it would need a 5,6 or 7-variable plot! ;-) Not simple
> variables like media method or number of samples, but color, density, color-map
> entries, etc.
Yes, obviously the ternary plot is only good for certain instances, but it might
be the best way to communicate the data in those specific instances.
Here's a 4-variable plot :D :D :D
http://udav.sourceforge.net/mgl_en.html/mgl_en_3.html
(scroll down to 3.6.9 Quaternary plot sample)
(I love all those cool-looking unusual types of graphing)
> But here's a simpler 3-variable example, with the sole purpose of making a
> decent-looking puffy 'sky cloud.' For the three variables, I'll choose only
> COLOR, DENSITY and TURBULENCE. Yet even with this 3-variable restriction, the
> resulting ternary plot would have only a tiny circular area somewhere within the
> triangle, representing 'a decent puffy cloud'. All other combinations of the 3
> variables would indeed produce *something*--but it wouldn't really look like a
> sky cloud.
Yes, but think about it this way - sometimes information on what NOT to do is
just as valuable as what to do.
"I want to experiment with media. Oh, look, Kenneth made a pretty graph of
different combinations ---- oh, all the combinations outside of that one little
region all suck. I won't bother wasting my time experimenting with those...."
[And I could _swear_ that someone once showed me a way to plot multiple
variables on 2d-graphs, much like the ternary plot, but I can't find any mention
of this or examples... Oh, brain, why do you invent these tantalizing false
memories....]
Post a reply to this message
|
|