Hands-On Unsupervised Learning with Python

A trade-off between homogeneity and completeness using the V-measure

The reader who's familiar with supervised learning should know the concept of F-score (or F-measure), which is the harmonic mean of precision and recall. The same kind of trade-off can be employed also when evaluating clustering results given the ground truth.

In fact, in many cases, it's helpful to have a single measure that takes into account both homogeneity and completeness. Such a result can be easily achieved using the V-measure (or V-score), which is defined as:

For the Breast Cancer Wisconsin dataset, the V-measure is as follows:

from sklearn.metrics import v_measure_score ...