How can I create a barseries plot using both grouped and stacked styles in MATLAB?

Question

The MATLAB bar documentation states the following:

bar(...,'style') specifies the style of the bars. 'style' is 'grouped' or 'stacked'. Default mode of display is 'grouped'.

However, I would like to achieve both at the same time. Let me elaborate by giving an example:

Y = [1.0 0.5 0.7
     2.0 1.5 2.0
     5.0 4.0 5.0
     4.0 4.0 4.5
     3.0 2.0 2.0];

bar(Y,'group');

This code produces the following grouped barseries plot, with 5 different sets of 3 bars grouped together:

bar([repmat(0.5,5,1) Y(:,1)-0.5],'stack');

And this code produces the following stacked barseries plot, using just the first column of the above defined matrix Y:

I would like to merge these two, to get a barseries plot which is grouped and stacked at the same time. So the desired result would be like the first picture and each of the three bars in a set would be stacked like the second picture.

Answer 1

There's no way I know of to get BAR to plot a combination grouped/stacked bar chart for you. However, you can do it yourself by adding rows of zeroes to your Y data wherever you want groups of bars to be separated, then modifying the x-axis tick marks and tick labels accordingly. Here's an example:

>> Y = [1 2 3; ...  %# Sample data
        2 3 4; ...
        3 4 5; ...
        4 5 6; ...
        5 6 7; ...
        6 7 8; ...
        7 8 9; ...
        8 9 10; ...
        9 10 11];
>> newY = reshape([reshape(Y,3,[]); zeros(1,numel(Y)/3)],[],3)  %# Add zeroes
                                                                %#   for spacing
newY =

     1     2     3
     2     3     4
     3     4     5
     0     0     0    %# <--- Note zero rows
     4     5     6
     5     6     7
     6     7     8
     0     0     0
     7     8     9
     8     9    10
     9    10    11
     0     0     0

>> bar(newY,'stacked');  %# Create a stacked histogram
>> set(gca,'XLim',[0 12],'XTick',2:4:10,'XTickLabel',1:3);  %# Modify axes

And here's the resulting figure:

Answer 2

group vs stacked is an either-or type of information. you need to be creative to do what you want to do:

bar(Y,'stacked','BarWidth',0.3);
hold;
X = Y.*abs(randn(5,3));
bar(X,'stacked','BarWidth',0.3, 'Xdata', 1.3:1:5.3);
Z = Y.*abs(randn(5,3));
bar(X,'stacked','BarWidth',0.3, 'Xdata', 0.7:1:4.7);

quick rundown: start off with stacked data, make sure the barwidth is small enough to fit data properly to the graph, hold your plot, then add the other bar plot with a data offset (Xdata).