How to reduce the file size of a saved histogram figure

In the following code, I have a histogram showing 100 bins, therefore the amount of data shown in this figure is quite small. However, the file saved is about 154 MB, and using the "compact" option it is still 77 MB.

Therefore, it seems the figure still contains the original data, which can indeed be seen using the Property inspector (see attachment).

Is it possible to save only the histogram data, and not the original data, such that the file saved has a minimal size?

x = randn(10^7,1);
h = figure;
histogram(x, 100)
savefig(h, "test.fig")
savefig(h, "test2.fig", "compact")

Answer 1

[Steven Lord]

You could avoid creating the histogram using the data by specifying 'BinCounts' and 'BinEdges'. If you do this, there won't be a way to retrieve the original data from the histogram the way you could if you created it by passing the unbinned data into the function, nor would you be able to manipulate it using functions like morebins or fewerbins or by manually changing some of the bin related properties.

Create a histogram from data

cd(tempdir)
x = randn(10^7,1);
f = figure;
h = histogram(x, 100);

saveas(f, 'FigureWithData.fig');

Create a histogram from counts and edges

f2 = figure;
[values, edges] = histcounts(x, 100);
h2 = histogram('BinCounts', values, 'BinEdges', edges);

saveas(f2, 'FigureWithoutData.fig');

Show the sizes of the files

D = dir('*.fig');
for thefile = 1:numel(D)
    fprintf("File %s has size %d.\n", D(thefile).name, D(thefile).bytes)
end

File FigureWithData.fig has size 154129315.
File FigureWithoutData.fig has size 27555.

Try manipulating the count-and-edges histogram

f3 = figure;
h3 = histogram('BinCounts', values, 'BinEdges', edges);

morebins(h3) % This will error

Error using matlab.graphics.chart.primitive.Histogram/morebins
morebins function is not supported when 'BinCountsMode' value is 'manual'.

[Jérôme]

Thank you for the detailed example. This is exactly what I wanted, I don't need the original data from the histogram, I just want the plot.

Answer 2

[Rik]

You have two options that I'm aware of:

1. Change the underlying data so you reproduce the same bins with fewer data points.
2. Change to a bar chart (use histcounts to extract the counts per bin).

[Jérôme]

The second option is a good one, thanks.