How can I extract a colormap from a dendrogram?

Hi, I have created a dendrogram and am happy with the colors it has selected by default. I want to create other plots using the same colormap to facilitate comparison. How do I extract the colormap from the dendrogram I have created?

Answer 1

[Jing Ci Neo]

Hello Rebecca,

I have a brute force solution for a few clusters. It looks like 'dendrogram' use the hsv colormap, and clusters with only 1 member are set to be black. The rest of the clusters are colored from

So what you can do is:

% Clustering
nClus = 10;                                         % Let's say you have 10 clusters
clustTreeEuc = linkage(eucD,'average');             % By average euclidean distance
color = clustTreeEuc(end-nClus+2,3)-eps;            % Calc. threshold based on nClus
[H,T,perm] = dendrogram(clustTreeEuc,0,'Orientation','right','ColorThreshold',...
    color,'labels','myDataNames');                  % Plot dendrogram
hidx = cluster(clustTreeEuc,'criterion','distance',...
    'MaxClust',nClus);                            % idx of data in each clus

% After you plot the dendrogram, find out what color each cluster is
% You can use histcounts(hidx) to help you
% By the order of the rainbow (r y g b ...) assign each cluster a new number
% For e.g. clus 1 is green, new no. is 3. Clus 2 is red, new no. is 1 etc.
% Black is the last color (clusters with 1 member)
cidx = [4 1 7 3 8 2 6 5 9 9];   % Example with two 1 member clusters (so 9 colors)

% Colors
cmap = hsv(max(cidx)-1);
cmap(end+1,:) = [0 0 0];
cmap(cidx(hidx(i)),:)       % This gives color of ith data

Hope it helps! If anyone finds a better method please let me know, thanks!

[Rebecca Ward]

Thankyou - this is really helpful. I've been doing it manually by selecting a line and inspecting the color which is OK for a small dendrogram but this is better :)

Answer 2

[Rishik Ramena]

You could use the ColorThreshold argument to specify the number of colors to be used in the dendrogram. This would then limit the use of newer colors in the different plots you seek to compare. You can also set the colors of the lines individually using the handles returned by the dendrogram function. Check here for examples and implementation.