On Thu, 24 Apr 2008, Dieter Menne wrote: > Achim Zeileis <Achim.Zeileis <at> wu-wien.ac.at> writes: > > However, I guess that it will be hard to select a qualitative > > palette with 18 distinct colors...I couldn't imagine a plot where it would > > be sufficiently easy for humans to decode that. But maybe you can combine > > that with some sequential or diverging palette or so? > > I slightly disagree here. In many cases, in color-coded surface plots, you do > not want to attribute the colors to, let's say, altitudes, but rather use the > color transitions as contour lines without explicitly drawing contours. You > could also use continuous colors, but it is amazing how this fails, while > something like 25 colors looks good.
I think it depends a lot on what you want to bring out in such a graphic. Qualitative palettes will convey the impression of different "groups" in the data so that distinction of various levels is easier but the impression of an underlying smooth curve is often lost. However, if you use a sequential palette, the smoothness is conveyed much better and it is easy to identify the "peaks" but much harder to compare other levels of your curve. (Re: my original comment. What I meant was that it would be challenging to decode some truly categorical data, i.e., without an underlying continuous scale, with 18 levels from a qualitative palette with 18 different colors.) Best, Z > Dieter > > ______________________________________________ > R-help@r-project.org mailing list > https://stat.ethz.ch/mailman/listinfo/r-help > PLEASE do read the posting guide http://www.R-project.org/posting-guide.html > and provide commented, minimal, self-contained, reproducible code. > > ______________________________________________ R-help@r-project.org mailing list https://stat.ethz.ch/mailman/listinfo/r-help PLEASE do read the posting guide http://www.R-project.org/posting-guide.html and provide commented, minimal, self-contained, reproducible code.
