Clustering of data using k-means

4 views (last 30 days)
Maria
Maria on 25 May 2018
Commented: darova on 6 Aug 2019
Hello, I have a vector with data and, when I plot them, they divide into four patterns, as you can see in the figure (the 2 of them in the middle are very close). I also know from the theory that I should have 4 of these curves. Because of numerical errors, I cannot analytically identify these groups. Therefore, I tried kmeans, and I used the following code
R = real(vector(1:2:end));
I = imag(vector(1:2:end));
X = [R I];
[idx,~] = kmeans(X,4,'replicates',5);%,'Distance','correlation');
figure
plot(X(:,1),X(:,2),'*')
%idx = clusterdata(X,'Maxclust',ncond,'Distance','jaccard');
figure;
plot(X(idx==1,1),X(idx==1,2),'r.','MarkerSize',12)
hold on
plot(X(idx==2,1),X(idx==2,2),'b.','MarkerSize',12)
hold on
plot(X(idx==3,1),X(idx==3,2),'k.','MarkerSize',12)
hold on
plot(X(idx==4,1),X(idx==4,2),'g.','MarkerSize',12)
legend('Cluster 1','Cluster 2','Cluster 3','Cluster 4')
However, how you can see in the figure, the clustering is horizontal, whereas it should be vertical.
I tried to tune the kmeans function with no success. I think the problem could be the different scale for the x and y data. I tried to multiply the x data for 1e2, but nothing changed. Does anyone have any idea how can I correctly cluster these data? I also attach the vector of data I am using. The values are complex conjugate, and in the plot, I am plotting only the values for the negative imaginary part.

Answers (0)

Products


Release

R2018a

Community Treasure Hunt

Find the treasures in MATLAB Central and discover how the community can help you!

Start Hunting!