How can I plot 18 lines in different colors on a single plot?

Hi everyone!

I have to plot 18 different functions of time, all in one single figure. How can i assign 18 different colors? (i work with matlab 2014)

Thanks.

Answer 1

[DGM]

In versions prior to R2019b, you can always set the axes property without the colororder() function.

% some fake data
x = linspace(0,1,100);
y = bsxfun(@plus,rand(18,100),(1:18).'); % i'm assuming we're in an older version

% set the axes colororder
CT = jet(18); % not that this is a good choice
set(gca,'colororder',CT);

% plot things
hold on
plot(x,y)

Of course, like @Bjorn Gustavsson says, unless you do something beyond this (e.g. using a carefully crafted map or using other line properties), putting 18 curves in the same plot can be a readability problem, especially if they overlap or the results need to be printed. For example:

clf
% exact same colors, but data series don't have sequential offsets
x = linspace(0,1,100);
os = (1:18).';
y = bsxfun(@plus,rand(18,100),os(randperm(18)));

% set the axes colororder
CT = jet(18); % not that this is a good choice
set(gca,'colororder',CT);

hold on
hp = plot(x,y);

% which curve is data 4?
legend(hp,'location','eastoutside')

See also:

https://www.mathworks.com/matlabcentral/fileexchange/70215-maximally-distinct-color-generator/

Answer 2

[Bjorn Gustavsson]

For that type of tasks I have had good use of the cmlines-function in the colormap-and-colorbar-utilities toolbox on the file exchange. With that it is as simple as:

ph = plot(randn(18,18));
colormap(turbo)
cmlines(ph(1:12))
colormap(copper)
cmlines(ph(13:18))

(unasked for advice: sometimes my plots become more legible if I also use linewith and linestyle to differentiate between lines)

HTH

Answer 3

[the cyclist]

If your version of MATLAB is 2019b or later, I suggest the colororder function. (I see that you mention an older version, so this may be more for other folks who come across this question.)

And for that many colors, I would recommend using a utility like this one for defining maximally distinguishable colors (and/or using different plotting symbols and line styles.)

[Bjorn Gustavsson]

Hopefully there are some more "natural" ordering and/or grouping of the function to plot, and then it might be way better to one distinct colormap per group of curves to guide the viewer.