The "best" number of clusters

You ran three different methods for finding the optimal number of clusters and their assignments and you arrived with three different answers.

Below you will find a comparison between the 3 clustering results (via coloring of the occupations based on the clusters to which they belong).

What can you say about the "best" way to cluster this data?