Scatter plot legend for only two of the three plotted categories

I have a 1738x6 matrix ("stock2"), from which I plotted the 6th column (y-axis) and the 5th column (x-axis). I divided values from the 6th column to three categories; top 100 (red dots), bottom 100 (blue dots) and the rest (green dots). I have extracted these high and low values, they are called "high100" and "low100" in the code below.

I understand that I have only one y-value in the plot and that it contains the three different categories. But I can't find a way to create a legend for the plot so that it would show only red dots and blue dots from inside my y-value. All attempts either fail or show a green dot and the first label of the legend. Could someone kindly show how to create the desired legend? Everything else in the code runs nicely. And as an extra question: why is there a [ ] in the scatter plot when using a color map?

figure
% color map
c = zeros(size(stock2,1),3);
middle = stock2;
[~,j] = sort(stock2(:,6),'ascend');
remove = j([1:100 end-99:end],:);
middle(remove,:)=[];
% other points are green so blue and red can be easily distinguished
% blue didn't seem to stand out from the default black dots
d=length(middle);
for i=1:d
    c(i,2)=1;
end
% red
a=length(middle)+1;
aa=a+99;
for i=a:aa
    c(i,1)=1;
end
% blue
b=length(middle)+length(high100)+1;
bb=b+99;
for i=b:bb
    c(i,3)=1;
end

scatter(stock2(:,5),[middle(:,6); stock2(high100,6); stock2(low100,6)],[],c,'.')
title('Stock2')
xlabel('Closing Price')
ylabel('Volume')
legend('100 highest volume days','100 lowest volume days')

Answer 1

[Voss]

scatter creates a single scatter object, so you won't be able to make multiple legend entries from it, but you can create additional lines to use in the legend (i.e., lines with NaN data so they don't show up in the plot, but they have the right properties - color, marker, etc. - to be used in the legend).

% random data:
stock2 = rand(1738,6);

figure

middle = stock2;
[~,j] = sort(stock2(:,6),'ascend');

% I imagine this is how high100 and low100 are defined:
high100 = j(end-99:end);
low100 = j(1:100);

remove = j([1:100 end-99:end],:);
middle(remove,:)=[];

% color map (simplified construction)
c = zeros(size(stock2,1),3);
n_middle = size(middle,1);
c(1:n_middle,2) = 1; % green
c(n_middle+(1:100),1) = 1; % red
c(n_middle+size(high100,1)+(1:100),3) = 1; % blue

scatter(stock2(:,5),[middle(:,6); stock2(high100,6); stock2(low100,6)],[],c,'.')
hold on
% plot two NaN lines for the legend
h = [ ...
    plot(NaN,NaN,'.','Color',c(end-100,:),'DisplayName','100 highest volume days') ... % red
    plot(NaN,NaN,'.','Color',c(end,:),'DisplayName','100 lowest volume days') ... % blue
    ];
legend(h);

title('Stock2')
xlabel('Closing Price')
ylabel('Volume')

"why is there a [ ] in the scatter plot when using a color map?"

The third input to scatter is the marker size, and you have to put something there if you want to use subsequent inputs, such as color and/or marker type. If you specify empty marker size, then MATLAB evidently uses the default value.

[Jonne Klockars]

Looks great, thank you very much! It's a bit over bedtime here so I'll run your solution tomorrow and after that accept your answer.

[Jonne Klockars]

'color' requires Simulink. I suppose there's no way of doing this without it?

[Voss]

Hmmm, check your code (or post it and I'll check it). My answer doesn't use a function called 'color'.

[Jonne Klockars]

My bad, a comment sign disappeared while I did a copy-paste with your code and some minor fixes to the style. ... is the " hold on " there because inside the legend two more plots are created?

[Voss]

"is the " hold on " there because inside the legend two more plots are created?"

That's essentially correct, yes.

But the two additional lines are not inside the legend; they are inside the same axes where the scatterplot is. Without hold on, the scatterplot would be deleted when the first additional line is plotted.

The handles of the additional lines (h) are used in the legend and the handle of the scatterplot is not.

[Jonne Klockars]

Ok, thank you very much for the lesson!

[Voss]

You're welcome!