V Score
The V-score or Normalized Mutual Information (NMI) score is a measure that combines the homogenity and completeness scores.
It is the harmonic mean of the two scores and defined by this formula.
This is similar to why we used the F1 score instead of just the recall and precision, we don't want our model to have a perfect homogenity score but no completeness and vice versa.
By optimising for the V-score we can train a model that is complete and homogenous.