Plot overlapped points (Matlab 2020a)

40 views (last 30 days)
I have a scatter plot (see the picture and code below) comparing the true values with the estimation results from 3 methods. The issue with this plot is that there are many overlapped points, making it hard to see.
My question is: How can we make the overlapped points to be seen? I am using Matlab 2020a. Thank you very much in advance !
Edit: Some suggested reducing the size of the points (Thank you for giving this suggestion). However, I actually do not want to reduce the Linewidth because there will be 4 plots like this one displayed in a figure. Also, I need to show all the points.
min_x = min(true_value(:));
max_x = max(true_value(:));
x = linspace(min_x -1,max_x+1,200);
hold on;
for k = 1:K
Estimates = mean(methods{k,1}.alpha,3);
set(gca,'fontsize',12) %
ylabel('True values');
legend({'method 1','method 2','method 3',},'Location','northwest','Interpreter',"latex");
hold off;

Accepted Answer

Meg Noah
Meg Noah on 28 Dec 2021
Just some ideas:
npts = 200;
true_value = randn(npts,1);
Estimates1 = true_value + 0.01*randn(npts,1);
Estimates2 = true_value + 0.08*randn(npts,1);
Estimates3 = true_value - 0.15*randn(npts,1);
s1 = scatter(Estimates1,true_value,20,'filled','DisplayName','Estimates1');
s1.MarkerFaceColor = '#0072BD';
s1.MarkerEdgeColor = 'None';
hold on
s2 = scatter(Estimates2,true_value,20,'filled','DisplayName','Estimates2');
s2.MarkerFaceColor = '#EDB120';
s2.MarkerEdgeColor = 'None';
s3 = scatter(Estimates3,true_value,20,'filled','DisplayName','Estimates3');
s3.MarkerFaceColor = '#A2142F';
s3.MarkerEdgeColor = 'None';
grid on
xlim([-4 4]);
ylim([-4 4]);
s1 = scatter3(Estimates1,3*ones(npts,1),true_value,'r','filled','SizeData',20);
hold on
s2 = scatter3(Estimates2,2*ones(npts,1),true_value,'y','filled','SizeData',20);
s3 = scatter3(Estimates3,1*ones(npts,1),true_value,'b','filled','SizeData',20);
grid on
xlim([-4 4]);
ylim([0 4]);
zlim([-4 4]);
  1 Comment
Hung Dao
Hung Dao on 28 Dec 2021
Thank you so much ! This is exactly what I am looking for.

Sign in to comment.

More Answers (2)

Sulaymon Eshkabilov
Sulaymon Eshkabilov on 27 Dec 2021
Use smaller line width within the loop, e.g.:
  1 Comment
Hung Dao
Hung Dao on 28 Dec 2021
Thank you for your suggestion. Is it possible to not reducing the linewidth?

Sign in to comment.

Image Analyst
Image Analyst on 28 Dec 2021
You probably don't need to plot so many points to get an idea of what's going on. Pick a subset of them, like 10 perent of the points or something. And reduce the LineWidth;
pct = 10;
numPointsToDisplay = round(pct * length(Estimates) / 100);
randomIndexes = randperm(length(Estimates), numPointsToDisplay);
plot(Estimates(randomIndexes), true_value(randomIndexes), shape{k}, 'LineWidth', 1);
Image Analyst
Image Analyst on 28 Dec 2021
If the line width is too big, many of the later-plotted markers will overlap and totally obscure the underlying markers so you won't see them anyway. Just look at your plot. The upper layer is just a solid mass of yellow. You're not seeing individual points, if that's what you (incorrectly) think.

Sign in to comment.


Community Treasure Hunt

Find the treasures in MATLAB Central and discover how the community can help you!

Start Hunting!