how to cluster a dataset which is extremly sparse(has missing values)


i have a dataset which is extremely sparse and i want to implement k-menas clustering ...

but k-means does ot work with missing values and i can not ommit missing values or replace them with a value or delete rows with missing values....any ideas?


one idea is, you could generate your own distance with the Java Distance node and use this together with the k-medoid node.

Best, Iris