Measures the separation distance between clusters. The Silhouette Coefficient s for a single sample is then given as:
- : mean distance between a sample and all other points in the same class
- : mean distance between a sample and all other points in the next nearest cluster
The Silhouette Coefficient for a set of samples is given as the mean of the Silhouette Coefficient for each sample. Score between -1 and 1 (-1 for incorrect clustering and +1 for highly dense clustering). The score is higher when clusters are dense and well separated. It can be used to select the optimal number of clusters.