Keeping same colors for the same groups in different group scatter plot

Hi!

I have a set of data containing around 5 000 000 different datapoints and these have been grouped into four different groups with the help of k-means clustering. When I plot these using gscatter, the four different colors presenting the datapoints belonging to each group in the plot are : group 1: purple, 2: blue, 3: orange and 4: yellow.

However, when I want to plot, for example, the first 100 000 of the datapoints they might only contain data belonging to group 1 and 3. Then the plot colours these as group 1: purple and group 3: blue instead of orange as in the original plot. How can I make it so the plots colours the datapoints based on the group they belong to?

Thank you in advance!

Answer 1

[KSSV]

P = rand(1000,2) ;
x = P(:,1) ; y = P(:,2) ;
idx = kmeans(P,4) ; 

figure
h = gscatter(x,y,idx) ; 

% Get colors
C = reshape([h.Color],3,[])' ; 
figure
hold on
scatter(x(idx==1),y(idx==1),[],C(1,:))
scatter(x(idx==2),y(idx==2),[],C(2,:))

[Oliver Wallin]

It does not seem to work if i do not want to plot every value that belongs to one group at the same time.

I have it like this:

x = data(:,1) ; y = data(:,2) ; z = data(:,3) ; w = data(:,4) ;
grp = kmeans(data,4);
% Then I divide all datapoints into colums with equal ammount of values
h = 0; 
for i = 1:length(data)
    g = 100*i;
    x1(:,i) = x(h+1:g);
    y1(:,i) = y(h+1:g);
    z1(:,i) = z(h+1:g);
    w1(:,i) = w(h+1:g);
    grp1(:,i) = grp(h+1:g); 
    h = 100*i
end

% Then I plot column by column for different variables, and this is one
% example

for j = 1:5
    figure
    gscatter(x1(:,j),y1(:,j),grp1(:,j))
end

So when I plot these values i want the colors to remain by group number by column. Because the first column might contain values from all four groups and the second column might contain from only 2 groups and so on.

[Oliver Wallin]

I solved that issue by creating new arrays in the plot loop. So now it works

Thanks for the help!