Random Clustering simply iterates over all examples and assigns a random cluster to it. Must be pure chance, if it performs better than k-Means, or your validation process is wrong.
As performance measurement I take the Item Distribution measurement. The other ones don't seem to work with clustering output, or only with one or two. I assument the measurement works ok, so I'm a bit puzzled.
Answers
Random Clustering simply iterates over all examples and assigns a random cluster to it. Must be pure chance, if it performs better than k-Means, or your validation process is wrong.
Best, Marius
As performance measurement I take the Item Distribution measurement. The other ones don't seem to work with clustering output, or only with one or two. I assument the measurement works ok, so I'm a bit puzzled.