Colours for different bar groups

Hello, I'm having problems to plot my data as a bar graph, with different colours for the different bar groups. I've read a bit about setting colours for bars, and that you would have to plot them individually to do so: https://se.mathworks.com/matlabcentral/answers/uploaded_files/59506/barchart.m

The problem is that I want to plot multiple bars grouped together in one plot, as automatically done when you have your data in a matrix. That looks like this and is almost the way I want it (but has the wrong colours):

Data(1,1)= 0.91;
Data(1,2)= 0.91;
Data(2,1)= 0.65;
Data(2,2)= 0.79;

barh(Data,1)
ylabel('gradient [permil]');
set(gca,'yticklabel',{'Control','Gas'});
set(gca,'YTickLabelRotation',90);

The first set is Data is duplicated because I actually have groups of different sizes and matlab doesn't allow that (but that's another problem). It would be fine for me to make one wide black bar out of this first set. So I would want to make the first two bars black, and the second two another colour, etc. (I will add more data). I tried doing this, but then the bars are plotted over one another:

mydata=Data;
figure(1)
hold on
a=1
for i = 1:length(mydata(a,:))
    h=barh(i,mydata(i,a));
    if i == 1
        set(h,'FaceColor','black');
    elseif 1 == 2
        set(h,'FaceColor','blue');
    else
        set(h,'FaceColor','red');
    end
    a=a+1;
end
hold off

This seperation into groups goes automatically in my first example, as I use a matrix. I just can't figure out how to do it when I plot them stepwise to get a handle on the colours. I hope you can help! Thank you

[Anne]

Hello, I figured out one way where I just plot a new bar over my old bars, which then also solves my problem with the different sizes of the groups.

h=barh(Data,1);
hold on
h(1).FaceColor = 'r';
h(2).FaceColor = 'b';
%here I plot my first bar over the wrongly-coloured first 2 bars
h=barh(Data(1,1));
h(1).FaceColor = 'k'; %now I can pick the color, yay!
hold off

If you know of a more elegant way, please tell me.

Answer 1

[Jos (10584)]

data = [0.91 NaN 0.65 0.79]
% the stacked trick
H = barh(1:numel(data), diag(data),'stacked') 
H(1).FaceColor = 'k' ; % set the colour of one bar

[Anne]

Hello Jos, thank you for your answer, but I don't understand it. How do I set the colour of the second bar, and how do I group bars (but not the first) together in different groups? Did you check my answer to myself, because it should look just like that?

Update: I think I see what you do, you create a grouped look by putting in NaNs in between. And because of your stacked command the bars somehow have handles so that I can change the colour. The how behind it is still a bit of a mystery to me though, so if you knw why this works, I would be curious to hear about it.

[Jos (10584)]

Hi Anne,

diag(data) creates a matrix where each column is plotted as a separate stacked bar. But because there are zeros in there, you don't see them

barh(diag(data)+1) % reveals the trick

There are some drawbacks of this too. You cannot set the width of an individual bar, for instance.

[Anne]

Thank you for the explanation!