Plot symbols the size of error

Hi folks,

I'm wondering if there's a way to plot a distribution with markers that are sized to correlate with your datas absolute uncertainty.

Say for example you have the following script:

x = rand(1,5)*10
err_x = rand(1,5)
y = rand(1,5)*10
err_y = rand(1,5)

figure()
errorbar(x, y, err_x, err_x, err_y, err_y)

I'm looking to replace the errorbar function with plot and a marker size that correlates to the size of the uncertainty.

I'd prefer not to plot the errorbars first and manually adjust the marker size. This method also requires a constant uncertainty across all values in a data set.

I'd greatly appreciate the insight.

Answer 1

[Star Strider]

I am not certain what you want as a result.

The scatter function is an option:

figure
scatter(x, y, err_x*100)
grid

If you want to plot the errors as ellipses:

a = linspace(0, 2*pi, 10);
elps = [cos(a); sin(a)];
for k = 1:numel(x)
    mrkrs(:,:,k) = [err_x(k).*elps(1,:)+x(k); err_y(k).*elps(2,:)+y(k)];
end
    
figure
scatter(x, y, '+')
hold all
for k = 1:size(mrkrs,3)
    plot(mrkrs(1,:,k), mrkrs(2,:,k))
end
hold off
grid
axis equal

Experiment to get the result you want.

[Lucas Warwaruk]

This is pretty intriguing. I think the scatter function will suffice. The ellipses are perfect for showing the unique errors between x and y dimensions.

Ultimately, I was hoping for a way to use the default markers in matlab and somehow size them to match the quantitative errors in different dimensions.

[Star Strider]

Thank you!

I am not aware that it is possible to alter the shapes of the built-in markers, only their sizes.