Computer scienceData scienceMachine learningClusteringClustering evaluation

Silhouette coefficient

The silhouette drawbacks

Report a typo

What are the most prominent drawbacks of using silhouette scores as the sole evaluator of clustering models?

Select one or more options from the list

Silhouette scores are hard to interpret

SIlhouette scores can't be used when there are many dimensions in the data

Silhouette can't be used to compare different algorithms with each other

Silhouette isn't suitable for clusters that have a non-spherical shape

___

Create a free account to access the full topic