Matlab：具有6 * 12矩阵颜色代码的3D饼图

Question

I was wondering if it was possible to color code a pie chart created on matlab with a different set of data than the one used to create the pie chart.

This chart is based on this code:

for F=1:6
    labels= {'Segment 1', 'Segment 2', 'Segment 3', 'Segment 4', 'Segment 5', 'Segment 6',         'Segment 7', 'Segment 8', 'Segment 9', 'Segment 10', 'Segment 11', 'Segment 12'};
    figure;
    x= [1 1 1 1 1 1 1 1 1 1 1 1];
    pie3(x,labels);
    colormap gray;
    title({'Floor n°=Figure n°';'Floor height=83.33mm';'Segment angle=30'});
end

What I want to do is to use a color code to assign values in the range 500 - 5000 from a 6*12 matrix to each segment of the pie chart(there are 6 of these figures, each showing 12 segments). Currently, the colormap has a range from 1-12 and gives a different color to each segment. I need the pie chart to remain the same as it represents a column divided into segments of equal volume. I am not a matlab expert but I was wondering if this was possible without having to create a pie chart from scratch, thus using the pie3 function.

Thanks in advance.

Answer 1

You definitely can color the pie charts (and 3D pie charts) with any color you like, and they do not need to be tied to the data initially used to create the pie chart.

Based on your explanations, I am not certain of what color you want to apply or how you are going to determine the color of a given segment of your pie, but I can give you an easy way to apply any color to a segment of a pie chart. This way you can decompose your problem into 3 steps:

• Generate your generic pie charts ( already done in your example code )
• calculate the colors of the segments, based on your algorithm
• Apply these color to the relevant segments of your pie charts ( where I come in )

---

How it works:

When you generate a 3D pie chart, Matlab first calculate the proportion to give to each segment ( all equals in your case ), then generates the graphic output. For that matlab generates four graphic objects per segment of the pie chart:

• 3 patch objects ( used to display the top, bottom and side of the given segment )
• 1 text object ( used for the text label of the segment )

The solution will simply consists in retrieving the handles of the graphic objects in order to assign a color to them.
Since in your case the collection of handle will be significant, we will also re-arrange it slightly to allow an easy allocation of color for a given segment, which will look like:

set( HandleCollection( FigureNumber , SegmentNumber) , desiredColor )

---

First, I add to modify slightly your example, because we need to retrieve the handles of the graphic objects at he time they are created (much easier this way). So here it is:

x= [1 1 1 1 1 1 1 1 1 1 1 1] ;

nPieChart = 3 ;         %// only 3 figures in this example, but any number can work
nSegments = length(x) ; %// number of segment for each pie chart

hPie = zeros( nSegments*4 , nPieChart ) ; %// initialise the handle matrix

% // Create your pie charts
for F=1:nPieChart
    labels= {'Segment 1', 'Segment 2', 'Segment 3', 'Segment 4', 'Segment 5', 'Segment 6',         'Segment 7', 'Segment 8', 'Segment 9', 'Segment 10', 'Segment 11', 'Segment 12'};
    figure;
    hPie(:,F) = pie3(x,labels) ;
    colormap gray;
    title({['Floor n°=' num2str(F)];'Floor height=83.33mm';'Segment angle=30'});
end

I took the definition of x out of the loop so I was able to pre-assign the size of the matrix hPie which will contain the handles (and also because if x never changes, no need to re-calculate it at every loop iteration. ( By the way, the same could apply to the labels if they do not change from one figure to another ).

Now we have a good collection of handle, let's reorder them in a more convenient manner. First we extract all the handles of the text labels ( we want these separated because they have different properties than the patch objects ):

idx_textHandles = 4:4:nSegments*4 ;
hLabels = hPie( idx_textHandles , : ).' ;

The last .' operator is used to transpose the matrix so we can acces the hLabels table by ( figureNumber , segmentNumber ). It seemed more intuitive to me to address the figure number before the segment number in an assignation .

Next we strip the hPie matrix of the text handles we just saved, then we reshape so the dimensions will be (m,n,p), with:

m = The number of figures
n = the number of segments in each pie chart
p = 3 (the 3 handles of the patch objects defining a segment)

hPie( idx_textHandles , : ) = [] ;
hSegment = permute( reshape( hPie , 3 , nSegments , nPieChart   ) , [3 2 1] ) ;

---

That's it ! You can now assign a color to a given segment with only one line of code, by setting the 'FaceColor' property of a patch object. For example the instruction:

set( hSegment( 2 , 5 , : ) , 'FaceColor','r' )

will color the segment #5 of the figure #2 in red. You can use any predefined color or the usual [RVB] triplets. You can also set the text of a given segment. So:

set( hLabels( 3 , 2) , 'String','HELLO')
set( hSegment( 3 , 2 , : ) , 'FaceColor', [.75 .75 .75] )

will color the segment #2 of the figure #3 in a light gray, and will set it's text label to 'HELLO'.

---

mmmh wait ! If you are keyboard lazy or simply if like me you are bothered with this type of matrix assignment (nFig, Nsegment, :) .After all, the last dimension of the matrix will always have to be assigned in full if we want to color the full segment, so having to specify the : every time is annoying ... No problem ... one more line of code and things will be even easier in the future:

hdlSegments = num2cell( hSegment, [nSegments  nPieChart] ) ;

Cool, we got rid of these trailing : in our assignments, now we can simply assign a color specifying the figure number then segment number. For example:

set( hdlSegments{ 3 , 6 } , 'FaceColor','m')

will set a nice magenta to the faces of the segment #6 of the figure #3.

Just pay attention now we have to use the {} instead of the () because we are accessing a cell array and not a simple numeric array anymore.

---

Not short enough ? extremely keyboard lazy ? ... ok last tip to reduce the syntax even more. One easy way would be to write a function to assign the 'facecolor' property of the 3 patch objects, but the code is so short it is almost a waste of a new file ... you can do it in one line:

colorSegment = @(fig,seg,color) set( hdlSegments{fig,seg} ,'FaceColor',color)

youhou, now you can type:

colorSegment( 3 , 4 , [0 0 1] )

and see the segment #4 of figure #3 change to a nice blue. Of course, if you did that because you are keyboard lazy you can give a shorter name to the function .