kmeans of 3d data (clustering)

I need to extend the clustering algorithm (Kmeans) to the third dimension. My dataset is composed: 700 row (different subjects) x 3 columns (each columns = different feature).

Is it possible to obtain the clustering graph in 3d? Is it possible to assign an index (for example 3 ouput, 1 2 3) for each subject based on the 3 characteristics?

For example, I want to assign to subject 1 (which corresponds three specific characteristics) a specific class belonging between one, secondo or third, (3=number of outputs defined a priori)

thanks a million

Answer 1

[Ameer Hamza]

kmeans() supports arbitrary numbers of features, so you can definitely use 3 features. Pass these features as a column of input matrix X. For example

X = rand(1000, 3); % 1000 samples with 3 features
idx = kmeans(X, 3); % 3 number of classes
scatter3(X(:,1), X(:,2), X(:,3), 15, idx, 'filled');

Answer 2

[Tara Rashnavadi]

this answer which uses plot3 rather than scatter3 worth reading too:

https://stackoverflow.com/questions/45152892/clustering-3d-scatter-data-using-k-means-clustering-in-matlab