Plotting multiple histograms in one figure

Hi,

I have some data points, simulated as follows:

for t=1:10000
H1(t)=normrnd(0,0.05);
H2(t)=normrnd(0,0.10); 
H3(t)=normrnd(0,0.30)
end

So essentially I generated three different random variables. I would like to do something very obvious.

I want to plot the histogram of each variable in the SAME graph/figure with the respective curve fitting to visualize the difference in the variances for these three random variables.

How I can implement this in Matlab?

Thanks

[Star Strider]

If you have the Statistics Toolbox, use the histfit function.

Answer 1

[dpb]

Another's shown the basics of adding to a plot; I'll note there's no need for loops and generating variables like H1, H2, H3 is generally bad practice in Matalab...use the vector facilities of Matlab, it is, after all, called "MATrix LABoratory" for a reason...

nSamp=10000;
mu=0;
sg=[0.05 0.1 0.3];
H=normrnd(mu,repmat(sg,nSamp,1));
hist(H,100), xlim([-1.3 1.3])

[Naser Zormati]

Does not work with UI axes! Is there another approach in this case?

[dpb]

The above was from what is now almost the dark ages in MATLAB version changes time frame... :)

For a uifigure, you'll have to add references to the figure and create the axes in it first...

nSamp=10000;
mu=0;
sg=[0.05 0.1 0.3];
H=normrnd(mu,repmat(sg,nSamp,1));
hUIF=uifigure;                   % create, save handle to uifigure
hAx=axes(hUIF);                  % create the axes in that figure, not default
hist(hAx,H,100)                  % plot into that axes, again not default
xlim(hAx,[-1.3 1.3])             % adjust limits of the specific axes

Since then, hist has been deprecated in favor of histogram; you probably should adjust to it although the above did work locally...

Answer 2

[Chad Greene]

After plotting the first histogram, you can use hold on to plot more histograms on top. If you're using Matlab 2014b or later, you can use the histogram function with 'facealpha' to set transparency. If you're using an older version of Matlab you can use histf in a similar fashion. I'm using 2012b here, with Stephen Cobeldick's brewermap function here:

map = brewermap(3,'Set1'); 

figure
histf(H1,-1.3:.01:1.3,'facecolor',map(1,:),'facealpha',.5,'edgecolor','none')
hold on
histf(H2,-1.3:.01:1.3,'facecolor',map(2,:),'facealpha',.5,'edgecolor','none')
histf(H3,-1.3:.01:1.3,'facecolor',map(3,:),'facealpha',.5,'edgecolor','none')
box off
axis tight
legalpha('H1','H2','H3','location','northwest')
legend boxoff

[msh]

That is great, but it could be ideal, If I can also have the fitting curve for each distribution with labels pointing the difference in variances. I am attaching a picture that I found where it shows exactly what I want, comparing three different distributions.

[Manuel Morgado]

Warning: histf() have some troubles in Matlab 2016 with legalpha() and rgb function.

For everything else, thank you Chad! :)

[Ninad Thakoor]

In the histf.m add following two lines after line 112 to fix alpha issues.

h.FaceAlpha = FaceAlpha;
h.EdgeAlpha = EdgeAlpha;

[siddharth rawat]

Thanks.

Answer 3

[Image Analyst]

Try something like

h1 = 0.05 * randn(1, 10000);
h2 = 0.10 * randn(1, 10000);
h3 = 0.30 * randn(1, 10000);
[counts1, binCenters1] = hist(h1, 500);
[counts2, binCenters2] = hist(h2, 500);
[counts3, binCenters3] = hist(h3, 500);

plot(binCenters1, counts1, 'r-');
hold on;
plot(binCenters2, counts2, 'g-');
plot(binCenters3, counts3, 'b-');
grid on;
% Put up legend.
legend1 = sprintf('mu = %.3f', mean(h1));
legend2 = sprintf('mu = %.3f', mean(h2));
legend3 = sprintf('mu = %.3f', mean(h3));
legend({legend1, legend2, legend3});