How to label a plot in Matlab with combine label?

It is hard to explaine here.

For example "peaks" is located second line and combine with "a" and "b".

Answer 1

[Adam Danz]

Option 1: use boxplotGroup()

See boxplotGroup() on the file exchange. If your boxplot data are matrices with the same number of columns, you can use this function to group them and provide primary and secondary labels.

Option 2: Use newline to create two rows of tick labels

Pro: Easy and clean

Con: Less control of the alignment between the upper and lower labels.

ax = axes(); 
xlim([0,12])
ticks = {'a_peaks','b','a_concat','b','a_F0','b','a_Fs','b'};
ticks = strrep(ticks,'_','\newline'); % replace underscore with \newline
ax.XTick = [1 2 4 5 7 8 10 11]; 
ax.XTickLabel = ticks; 

This variation below controls the alignment between the upper and lower labels but adds an additional tick between the upper labels.

ax = axes(); 
xlim([0,12])
ticks = {'a','_peaks','b','a','_concat','b','a','_F0','b','a','_Fs','b'};
ticks = strrep(ticks,'_','\newline'); % replace underscore with \newline
ax.XTick = [1 1.5 2   4 4.5 5   7 7.5 8   10 10.5 11]; 
ax.XTickLabel = ticks; 

Option 2.1: Use cell array with justified text

Added to answer 11-Feb-2021

Pro: Intuitive label arrangement & text justification

ax = axes(); 
xlim([0,12])
ticks = {'a','','b','a','','b','a','','b','a','','b'; ...    % 1st row of labels
    '','peaks','','','concat','','','F0','','','Fs',''};  % 2nd row of labels
ticksJust = strjust(pad(ticks),'center'); % Justify: 'left'(default)|'right'|'center

% ticksJust =
%   2×9 cell array
%     {'  a  '} {'     '} {'  b  '} {'  a  '} {'     '} {'  b  '} {'  a  '} {'     '} {'  b  '}
%     {'     '} {'peaks'} {'     '} {'     '} {'peaks'} {'     '} {'     '} {'peaks'} {'     '}

tickLabels = strtrim(sprintf('%s\\newline%s\n', ticksJust{:}));
ax.XTick = [1 1.5 2   4 4.5 5   7 7.5 8   10 10.5 11]; 
ax.XTickLabel = tickLabels; 

Option 3: Use text() to create a second row of tick lables

Pro: More control over the placement of the tick labels.

Con: Axis limits must be set and the lower tick labels will move if the pan feature is used.

ax = axes(); 

% Set axis limits before labeling
xlim([0,12])
ylim([0,1]); 

% Shorten the height of the axes to 90% and shift  
% upward to make room for the 2-layers of labels.
upperPosition = sum(ax.Position([2,4])); 
ax.Position(4) = ax.Position(4) * .9; 
ax.Position(2) = minus(upperPosition, ax.Position(4)); 

% label the upper row using xtick and xticklabel
ax.XTick = [1 2 4 5 7 8 10 11]; 
ax.XTickLabel = {'a' 'b'}; 

% use text() to creat the lower row of tick labels
% The 2nd layer will be 6% of the y axis range below the axis.
lowerLevel = min(ylim(ax))-range(ylim(ax))*.06; 
lowerLabels = {'peaks','concat','F0','FS'}; 
text([1.5, 4.5, 7.5, 10.5], repmat(lowerLevel,1,numel(lowerLabels)), lowerLabels,...
    'VerticalAlignment','Top','HorizontalAlignment','Center')

[Matlaber]

Thanks. Can i ask if this applied to boxplot function?

I tried, but it not working well.

[Adam Danz]

If you share the code you tried, I could help fix it.

[Matlaber]

Thanks for your reply.

I not sure how to add in boxplot function

boxplot(rand(100,2),{'a','b'})
xlabel('Peaks','FontSize',12)

[Adam Danz]

With this syntax, you only need to apply the 2nd row of labels since the first row is provided by the boxplot() function.

For example,

ax = gca(); 
lowerLevel = min(ylim(ax))-range(ylim(ax))*.06;
lowerLabels = {'peaks'}; % <--- define lower labels here
lowerLabelsX = [1.5];    % <--- define the x-position of each label
text(lowerLabelsX, repmat(lowerLevel,1,numel(lowerLabels)), lowerLabels,...
    'VerticalAlignment','Top','HorizontalAlignment','Center')

[Adam Danz]

Note that I've updated my answer with a new Option 1.

See the boxplotGroup() function on the file exchange

https://www.mathworks.com/matlabcentral/fileexchange/74437-boxplotgroup

[Matlaber]

Thanks. Your answer very good.

Can I ask how to lable 4 values like below:

boxplot(randn(4),{'a','b','c','d'})
xlabel('Peaks', 'Peaks2','FontSize',12)

[Adam Danz]

To use the boxplotGroup() function, you just need to split the matrix up into a cell array which is demonstrated below.

data = randn(4); 
splitData = {data(:,1:2), data(:,3:4)}; 
% or
% splitData = mat2cell(data, size(data,1), [2,2]);

boxplotGroup(splitData,'primaryLabels', {'a' 'b' 'c' 'd'}, 'SecondaryLabels', {'Peak', 'Peak1'})

If you're rather not have the horizontal space between groups, set "InterGroupSpace" to 0.

Also see this link for more examples.

[Matlaber]

Thanks, the function you created is brillant!

Between, if I add a y-axis, it seemed overlap with the y-axis label.

[Adam Danz]

I didn't understand that last part about adding a y axis. Can you share a screen shot and a snippet of relevant code?

[Matlaber]

Sorry for my bad English

data = randn(4); 
splitData = {data(:,1:2), data(:,3:4)}; 
% or
% splitData = mat2cell(data, size(data,1), [2,2]);
boxplotGroup(splitData,'primaryLabels', {'a' 'b' 'c' 'd'}, 'SecondaryLabels', {'Peak11', 'Peak1'})
ylabel('Percentages')

[Adam Danz]

Thanks for bringing that to my attention. I fixed that in the boxplotGroup() function and updated the file on the file exchange (version 1.2.0).

You can download the new version and the problem should be fixed.

[Matlaber]

I should thanks you.

However, could you check again? I am not sure is my file loaded wrongly.

But the new code, y-axis does not show up.

data = randn(4); 
splitData = {data(:,1:2), data(:,3:4)}; 
% or
% splitData = mat2cell(data, size(data,1), [2,2]);
boxplotGroup(splitData,'primaryLabels', {'a' 'b' 'c' 'd'}, 'SecondaryLabels', {'Peak11', 'Peak1'})
ylabel('Percentages')

[Adam Danz]

It looks fine to me (screen shot below). What version of Matlab are you using? Are you sure you're using the newest version of boxplotGroup()?

[Matlaber]

Hi, Thanks.

I am usng MATLAB R2017b and also I using the newest version of code.

Is below will give some hints why not working?

>> data = randn(4); 
splitData = {data(:,1:2), data(:,3:4)}; 
% or
% splitData = mat2cell(data, size(data,1), [2,2]);
boxplotGroup(splitData,'primaryLabels', {'a' 'b' 'c' 'd'}, 'SecondaryLabels', {'Peak11', 'Peak1'})
ylabel('Percentages')

ans = 

  LinkProp with properties:

          Enabled: 'on'
    PropertyNames: {'Units'  'Position'  'ActivePositionProperty'  'Parent'}
          Targets: [1×2 Axes]

No appropriate method, property, or field 'Toolbar' for class 'matlab.graphics.axis.Axes'.

Error in boxplotGroup (line 172)
        h.axis2.Toolbar.Visible = 'off';
 

[Adam Danz]

ah.... there is a version issue between r2017b and r2019b. Thanks again for bringing this to my attention.

I've update the file and you can download the most recent version 1.2.1. I've tested it on r2017b and r2016b.

[Matlaber]

Thanks, it working very well.

Just one question. Regarding the range of boxplot between the function of "boxplot" and "boxplotGroup".

For example:

Using MALTAB function "boxplot", the y-axis start from 30 to 100.

Using "boxplotGroup" function, the y-axis start from 0 to 100.

%%
y = [33, 44, 67, 78, 80, 90, 99, 47];
boxplot(y,'Labels',{'1'})

%%
boxplotGroup(y)

Error using boxplotGroup
The value of 'x' is invalid. Expected input to be one of these types:

cell

Instead its type was double.

Error in boxplotGroup (line 95)
parse(p,varargin{:})

Error in Boxplot2 (line 129)
boxplotGroup(x)

Thanks.

[Adam Danz]

If you execute

help('boxplotGroup')

you'll notice that the 'x' input should be a cell array.

%   boxplotGroup(x) receives a 1xm cell array where each element is a matrix with
%   n columns and produced n groups of boxplot boxes with m boxes per group.

Here's the correct way to set that up.

figure()
y = [33, 44, 67, 78, 80, 90, 99, 47];
ax1 = subplot(1,2,1);
boxplot(y); 
ax2 = subplot(1,2,2);
boxplotGroup(ax2, {y(:)})
xlim(ax2, xlim(ax1)) % This just equates the scaling

Regarding the difference in the y-axis and x-axis limits between boxplot() and boxplotGroup(), that is expected. The limits are set within the boxplotGroup() function but you can change them as needed after the plots are produced.

[Matlaber]

Thanks for your reply.

It is still a bit difficult for me. Not sure why still not working with the y-axis limit.

1. Using "boxplotGroup" function

a= [33.3333  100.0000   61.9048   95.2381;
   57.1429   80.9524   66.6667  100.0000;
   57.1429   85.7143   47.6190   95.2381;
   42.8571   90.4762   52.3810   80.9524;
   42.8571   95.2381   61.9048   85.7143;
   57.1429  100.0000   71.4286  100.0000;
   42.8571  100.0000   28.5714   85.7143;
   61.9048   95.2381   76.1900   66.6667;
   38.1000  100.0000   57.1429   76.1905;
   71.4286   90.4762   52.3810  100.0000]
splitData = {a(:,1:2), data(:,3:4)}; 
boxplotGroup(splitData,'primaryLabels', {'a' 'b' 'c' 'd'}, 'SecondaryLabels', {'Range1', 'Range2'})
ylabel('Percentages')

Output:

2. Using function "boxplot"

a= [33.3333  100.0000   61.9048   95.2381;
   57.1429   80.9524   66.6667  100.0000;
   57.1429   85.7143   47.6190   95.2381;
   42.8571   90.4762   52.3810   80.9524;
   42.8571   95.2381   61.9048   85.7143;
   57.1429  100.0000   71.4286  100.0000;
   42.8571  100.0000   28.5714   85.7143;
   61.9048   95.2381   76.1900   66.6667;
   38.1000  100.0000   57.1429   76.1905;
   71.4286   90.4762   52.3810  100.0000]
boxplot(a,'Labels',{'AM','b','a','c'}) 

Thanks.

[Adam Danz]

The only significant difference in the figures you shared is the order the boxplots are in. See the inline comments below to produce an exact replication between the two functions.

% Input data
a= [33.3333  100.0000   61.9048   95.2381;
   57.1429   80.9524   66.6667  100.0000;
   57.1429   85.7143   47.6190   95.2381;
   42.8571   90.4762   52.3810   80.9524;
   42.8571   95.2381   61.9048   85.7143;
   57.1429  100.0000   71.4286  100.0000;
   42.8571  100.0000   28.5714   85.7143;
   61.9048   95.2381   76.1900   66.6667;
   38.1000  100.0000   57.1429   76.1905;
   71.4286   90.4762   52.3810  100.0000]

% ******* boxplotGroup setup ******* 
% Note that the columns of 'a' are rearranged in the line below. 
splitData = {a(:,[1,3]), a(:,[2,4])}; 
% Plot results
figure()
boxplotGroup(splitData,'primaryLabels', {'a' 'b' 'c' 'd'}, ...
    'SecondaryLabels', {'Range1', 'Range2'}, 'interGroupSpace', 0)
%                           this section here ^^^^^^^^^^^^^^^^^^^^
%                           removes the space between the box groups
ylabel('Percentages')
grid on
ylim([25, 104])

% ******* boxplot setup *******
figure()
boxplot(a,'Labels',{'a' 'b' 'c' 'd'}) 
grid on
ylim([25, 104])

[Matlaber]

Thanks. Brillant!

Is there a way to increase the label font size?

With this below, only the first line font increase.

set(gca,'FontSize',13)

[Adam Danz]

Use the outputs

h = boxplotGroup(. . . ); 
set(h.axis,'FontSize',13)
set(h.axis2,'FontSize',13)

[Matlaber]

It works well. Thanks.

[Adam Danz]

Glad I could help. Thanks for testing the boxplotGroup function and providing feedback.

[Matlaber]

my pleasure. I also leave a good comment on your function too.

[Adam Danz]

Thank you!!

[Matlaber]

Will it be possible to make one of the words in tha label in italic font?

From

boxplot(a,'Labels',{'a' 'b' 'c' 'd'}) 

to

boxplot(a,'Labels',{\it'a' 'b' 'c' 'd'})    

[Adam Danz]

ax.XAxis.FontAngle = 'italic';

where ax is your axis handle.

[Adam Danz]

@Hsein Ping Kew

If you want 2 groups with 3 boxplots per group, your input will be a 1x3 cell array.

Each of the 3 cells will should contain 2 columns where column 1 are the data for the first boxplot of each group, column 2 are the data for the 2nd boxplot for each group

splitdata = {c(:,[1,4]),c(:,[2,5]),c(:,[3,6])};

Please let me know if you have any problems or need more clarification.

Answer 2

[Akira Agata]

I believe the simple way is to separate the plot for each group and use xlabel to add 'Peaks', 'Concat', etc.

The following is an example:

figure
subplot(1,2,1)
boxplot(rand(100,2),{'a','b'})
xlabel('Peaks','FontSize',12)
subplot(1,2,2)
boxplot(rand(100,2),{'a','b'})
xlabel('Concat','FontSize',12)

[Matlaber]

Thanks. Is there a away to combine both boxplot together?