Combining legend data

What I would like to do is combine two legend items together. Suppose I have some like the following:

plot(4.5+rand(1,50),'o')
hold on
plot(5*ones(1,50))
legend('data','fit')

But instead of having 2 entries in the legend (1 circles, 1 line) I have only 1 entry (line with circle in the middle).

Does anybody know how to do this?

Answer 1

[Martijn]

What you could do here is first change your code a little to obtain handles to the plot:

p1= plot(4.5+rand(1,50),'o');
hold on
p2 = plot(5*ones(1,50));

Then only add a legend to the second plot (again make sure you obtain a handle):

l = legend(p2,'MyLegend');

Then modify the Marker used in this legend:

c = get(l,'Children');
set(c(1),'Marker','o')

[Steven White]

Perfect! This does exactly what I want it to do.

Many thanks,

Steven.

[Yogesh T L]

I have experimental data and fitted model with markers and lines respectively. So, I also have a similar situation like Stevan and tried to apply the above code to combine the markers and line in the legend but it doesn't work for me. So, please let me know if you have any suggestions.

[Martijn]

The code above is for MATLAB releases before R2014b. For MATLAB release R2014b and newer with the new graphics system, this would become:

p1= plot(4.5+rand(1,50),'ob');
hold on
p2 = plot(5*ones(1,50),'b');
[~,objects] = legend('MyLegend');
objects(2).LineStyle = '-';

[Jeen Doe]

In 2017a at least, the last line should be

objects(2).Children.LineStyle = '-';

Answer 2

[Walter Roberson]

When I try in 2008b Linux 64, I get two legend entries, as I expect.

Are you finding that you only get one legend entry, or are you trying to get only one legend entry? If you are trying to get only one legend entry, what are you hoping that it would look like considering you have two labels and two markers?

[Steven White]

I would only like one entry. The idea here is to represent the circle and the line as the same thing. For example, this could be data (circles) and fitted lines (lines). Hence, if I am plotting multiple datasets and fitted lines, 4 of each say, then I only have 4 legend entries, rather than 8.